Discussion

Revision surgery for chronic otitis media ideally results in a manageable and dry middle-ear cavity, but it is often plagued with high failure rates, with cholesteatoma recurrences found in up to 49 per cent of patients after previous cholesteatoma surgery.^{Reference Nadol6,Reference Veldman and Braunius7} Recurrent disease is commonly identified when initial surgery fails to address tegmental, perifacial, mastoid apex and sinodural mastoid cells.^{Reference Kasenõmm11,Reference Košec, Kelava, Ajduk, Ries, Trotic and Bedekovic12} However, clinical follow up using otoendoscopy and presenting symptoms may not identify a significant portion of patients with occult disease recurrence, especially after previous canal wall up surgery.^{Reference Yu, See, Ng, Low, Tan and Yuen8,Reference Campos, Mata, Reboll, Peris and Basterra9}

Pre-operative CT imaging is a first-line option in middle-ear imaging because of its excellent spatial resolution and delineation of vital landmarks.^{Reference Campos, Mata, Reboll, Peris and Basterra9,Reference Coralles and Blevins10,Reference Gul, Akdag, Kinis, Yilmaz, Sengul and Teke14} It is especially useful in cases where the initial surgery was performed elsewhere, if complications are present, or when recurring disease cannot be properly assessed using otoendoscopy alone.^{Reference Coralles and Blevins10} It is used to predict the extension of the disease into the sinus tympani and the attic, with antral and peri-antral involvement.^{Reference Campos, Mata, Reboll, Peris and Basterra9} However, when interpreting CT scans, the inability to discriminate between cholesteatoma, granulation, fibrosis and thick mucus secretion in the middle ear must be taken into account; one should consider performing magnetic resonance imaging (MRI) as well, especially in the paediatric population.^{Reference Campos, Mata, Reboll, Peris and Basterra9,Reference Romano, Covelli, Confaloni, Rossi-Espagnet, Butera and Barbara15} Recurrent middle-ear cholesteatoma evaluated by MRI also represents a diagnostic challenge because of the small size of these tissues, the different kinds of surgical procedures, and the non-specific signal intensity characteristics on imaging. The presence of erosion caused by soft tissue on nearby bone structures is generally interpreted as indicating the presence of cholesteatoma.^{Reference Gul, Akdag, Kinis, Yilmaz, Sengul and Teke14,Reference Romano, Covelli, Confaloni, Rossi-Espagnet, Butera and Barbara15}

Previously published studies have been concerned with comparing pre- and post-operative CT and MRI scans in relation to initial surgery for chronic otitis media. However, very few studies have addressed the much more complicated issue of whether pre-operative imaging is useful in revision surgery. While MRI is still widely regarded as the most accurate imaging method for assessing cholesteatoma, high-resolution CT may represent an alternative imaging tool in patients with contraindications for MRI or in practices where pre-operative MRI is not routinely performed. At the cost of a low radiation burden, CT represents a cheaper and faster examination, with the additional advantage of providing high-resolution anatomical details.^{Reference Lingam and Bassett16,Reference Khemani, Singh, Lingam and Kalan17} Advancing imaging protocols may further minimise radiation, and, in some instances, cone beam CT may be useful. However, cone beam CT may not capture areas of interest outside the field of view (beyond 4 × 4 cm classically).^{Reference Stutzki, Jahns, Mandapathil, Diogo, Werner and Güldner18} Cone beam CT may aid the surgeon's pre-operative plan and discussion of treatment options, which may be different from decisions made based on intra-operative findings or pre-operative imaging. In revision cases, the possibility of intra-operative complications and plan changes is higher, and the surgeon may be required to widen the extent of surgery beyond the original plan.

Several studies have assessed the value of diffusion-weighted MRI and concluded that a high signal on diffusion-weighted images does not always indicate the presence of cholesteatoma, with false-positive cases found in the post-operative period due to artefacts, bone dust, cartilage, tympanosclerosis, granulation tissue, proteinaceous fluid, purulent content, cholesterol granuloma or wax in the adjacent external auditory canal.^{Reference Romano, Covelli, Confaloni, Rossi-Espagnet, Butera and Barbara15} A recent meta-analysis of 1152 patients (both pre- and post-surgical cases) showed that the use of diffusion-weighted images alone led to 3.4 per cent of false-positive cases, with a lower specificity for post-operative controls.^{Reference Lingam and Bassett16}

The principal question analysed in this study was whether pre-operative CT scans may present a useful tool in identifying localisations with high risk for recurrent disease, and in planning for revision surgery. The results confirm our hypothesis that CT can reliably identify recurrent disease. In the current study, CT identified 25 out of 31 cholesteatoma recurrences, with high sensitivity (80.65 per cent) and a high positive predictive value (92.59 per cent). This value is much higher than some of the results in the literature, which cite sensitivity and specificity as low as 42 per cent and 48 per cent, respectively.^{Reference Khemani, Singh, Lingam and Kalan17,Reference Songu, Altay, Onal, Arslanoglu, Balci and Ucar19,Reference Alzahrani, Alhazmi, Bélair and Saliba20} Pre-operative CT scans revealed correlations between the presence of cholesteatoma and attic disease, ossicular chain involvement, sinodural angle cells, opacified facial ridge cells and an obstructed antrum. In addition, the likelihood of performing canal wall down tympanomastoidectomy was higher in patients presenting with disease in the sinodural angle, erosion of the posterior canal wall and exposed dura on the pre-operative CT scan, which was further confirmed by principal component analysis results.

Male sex was linked to a higher risk of cholesteatoma recurrence, which has been described earlier, and indicates that our sample is representative of the larger patient population.^{Reference Allam, Abdel Razek, Ashraf and Khalek21}

The most common localisations of recurrent disease visible on pre-operative CT were tympanic perforation, ossicular chain involvement, posterior canal wall destruction, attic disease and antrum involvement. This correlated well with intra-operative findings (Table 3). These findings correspond to well-known disease recurrence patterns in published literature, confirming that failure to address these areas during initial or previous surgery may be a principal factor in disease recurrence.^{Reference Allam, Abdel Razek, Ashraf and Khalek21}

The results of the matched pair analysis also indicate that performing pre-operative CT scans increases the rate of accurate diagnosis of recurrent cholesteatoma, reduces the time required to diagnose recurrent disease, and lowers the average age at diagnosis and corresponding revision surgery, without influencing the type of surgery. Analysis of correlations between pre-operative CT scans and intra-operative findings in revision surgery is clearly of value when planning the extent and type of surgery, and when determining which areas are high risk for subsequent close scrutiny, without introducing bias when choosing the type and extent of surgery required to eliminate recurrent disease.^{Reference Alzahrani, Alhazmi, Bélair and Saliba20}

The habit of analysing pre-operative imaging must be a valuable tool in any surgeon's armamentarium, and should not be relegated exclusively to the radiologist. However, a capable surgeon should be able to navigate a temporal bone to ensure that the bone bed is clear and comprehensively drilled as part of any approach when removing cholesteatoma disease, irrespective of pre-operative scanning. The point of obtaining the CT scan prior to surgery is not to aid the surgeon in identifying areas in the ear where there is concern for recurrence. It might seem obvious that CT may help with surgical planning, but data supporting objective analysis of the CT scan in revision surgery are surprisingly scarce, and supporting its consistent use would help clinicians show value to third parties who may question the need for CT in revision cases. Knowledge of fine anatomical detail provides additional refinement in surgical technique, which is a decisive factor influencing post-operative outcomes.^{Reference Košec, Kelava, Ajduk, Ries, Trotic and Bedekovic12,Reference Faramarzi, Motasaddi-Zarandy and Khorsandi13} Recurrent cholesteatoma is associated with inaccurate pre- and intra-operative evaluation findings of the surgical field and a failure to address high-risk areas, putting patients at a higher risk of canal wall down procedures in revision surgery.^{Reference Košec, Kelava, Ajduk, Ries, Trotic and Bedekovic12} The correlations between CT scans and intra-operative findings identified in this study may be useful in follow up as well, as indicators of early recurrence, and for prompt timely surgical intervention in patients with otherwise occult disease.^{Reference Allam, Abdel Razek, Ashraf and Khalek21}

When planning for surgery, it may be especially valuable to have in mind the significant correlations between disease involving the sinodural angle and posterior canal wall and the presence of exposed dura, as early indicators that a canal wall down procedure might be required (Table 5). Similar findings have been reported previously, with intra-operative findings identifying posterior external auditory canal wall erosion, mastoid erosion and residual cells along the facial ridge as reliable predictors of cholesteatoma recurrence, and associated with canal wall down procedures, but they have not been linked with pre-operative CT findings.^{Reference Kasenõmm11–Reference Faramarzi, Motasaddi-Zarandy and Khorsandi13,Reference Selwyn, Howard and Cuddihy22}