What to report in sellar tumor MRI? A nationwide survey among German pituitary surgeons, radiation oncologists, and endocrinologists

Pituitary fossa and craniopharyngeal duct tumors account for 16.6% of all primary brain and central nervous system tumors, according to the Central Brain Tumor Registry of the United States (CBTRUS) statistical report 2008–2012 [1]. Most of these pathologies are histologically benign, with pituitary adenomas representing the most common entities [1, 2]. While computed tomography is still useful in assessing the osseous anatomical conditions of the skull base and tumor calcifications, it has been largely replaced by magnetic resonance imaging (MRI) as the imaging modality of choice in the evaluation of sellar masses due to its higher soft tissue contrast [3, 4]. With steady development of imaging techniques and equipment including high-field 3-T MRI, anatomical relationships and pathologies can be described with increasing detail. Based on these prerequisites, experts reported recommendations both for the execution of MRI examinations of hypophyseal fossa tumors and for the content of respective radiology reports [4‐7]. To the best of our knowledge, however, no attempt has been made so far to correlate the content of radiology reports with the information needed by the interdisciplinary team dealing with sellar tumor patients, although such a single-center initiative has already resulted in increased satisfaction among referring physicians in the case of primary brain tumors at Emory University School of Medicine, Atlanta, Georgia, USA [8, 9]. Therefore, we aimed to determine the specific content preferences for MRI reports of neoplastic lesions of the sellar region by board certified neurosurgeons, radiation oncologists, and endocrinologists in a nationwide survey.

Through the extensive survey presented in this study, we explore for the first time on a national scale the demands that treating clinicians place on MRI reports in the context of caring for patients with tumors of the sellar region. Although several recommendations have already been published in the past by experienced neuroradiologists on the realization and reporting of MRI scans for the evaluation of sellar masses [6, 7, 11‐13], only very few of them are based on a consensus or on a systematic survey and analysis of the information needs and expectations of the treating referring specialists who are the primary target audience for the MRI reports [14, 15]. In fact, it appears that to date the desire for standardized style and billing considerations influence the development of structured radiology reporting to a far greater extent than do the requirements of referrers, as a national survey of academic radiologists in the United States of America showed [16]. While this article primarily focuses on the provision of high quality radiology reports that meet the information needs of clinicians, it should not be ignored that this can only be achieved by the radiologist through a sufficient information base in the form of an adequate patient history and justification for imaging by the referring clinician [17]. Whereas in modern market economies, the wishes of consumers or customers play a central role in the production of a commodity or the provision of a service; few attempts have been made by radiological protagonists to determine the satisfaction of referring physicians as customers of the product “MRI report” or to optimize the radiological report regarding existing needs [16]. Only a small number of studies have so far taken stock of the situation and revealed deficits in the reporting from the point of view of the addressees. For example, a survey of general practitioners in the United Kingdom revealed that the majority were not familiar with the normal size ranges of frequently measured and reported anatomical structures, and a survey of medical oncologists in Australia revealed that key lesion sizes, which they considered very important, were often not available and that a desired comparison with previous examinations was often not carried out [18, 19]. Boll and colleagues took a structured approach to this problem and, using the so-called “voice-of-the-customer method,” were able to show at their institution that the greatest deficits perceived by clinicians in radiology reports were insufficient consideration of their specific information needs and the lack of communication of key information relevant to practice [20]. This feedback should be considered by clinical radiologists as a valuable tool with a potential for qualitative and quantitative enhancement of their reporting activities. The basic idea of a systematic orientation toward the specific information needs of referring clinicians has so far been taken up in the neuroradiological-oncological context only for primary brain tumors (gliomas), on the one hand, monocentrically at Emory University (Atlanta/USA) and, on the other hand, in the context of a nationwide survey among neurosurgeons, radiation therapists, medical oncologists, and neuropathologists in Germany [9, 21]. Implementation of the so-called “brain tumor reporting and data system (BT-RADS)” resulted in higher satisfaction among clinicians with radiologic image analysis through improved coherence, unambiguity, and interdisciplinary communication in a follow-up of the first-mentioned study after the MRI reporting elements were adapted to the preferences of referring physicians [8]. In addition, it has already been shown in other subdisciplines that radiologists also prefer department-wide standardized structured reporting after its pervasive implementation in the longer term. According to Larson et al., this could only be achieved by closely involving all medical staff of the radiology department in the initial template creation process, avoidance of excessive restriction of reporting by a certain degree of flexibility in describing abnormal findings, and continuously considering user feedback after the introduction of the templates, since there was initially a certain skepticism due to the anticipated loss of autonomy [22]. Following this guiding principle, we present here an evaluation of potential MRI reporting elements for patients with tumors of the sellar region, which could be collected for the first time through a nationwide survey among clinical experts in the disciplines of neurosurgery, radiotherapy, and endocrinology relevant to the management of these conditions [23]. Our survey results are largely consistent with the consensus-based proposal of a small ENT- and skull base surgery-oriented expert group (3 neuroradiologists, 3 ENT specialists, and 3 skull base surgeons) from Melbourne for a structured pituitary MRI reporting template, which also includes adenoma size, internal composition of the lesion, relationship to pituitary tissue/infundibulum, and impairment of adjacent structures (optic chiasm, cavernous sinus, and internal carotid artery) [14]. This template lists detailed characterization of the sphenoid sinus (size and pneumatization) and the nasal cavity (septal deviation, Onodi cells, and changes due to previous sinonasal surgery) as essential additional reporting elements. Although these items were rated as important in the overall evaluation in our study by only a minority, they received significant agreement among the neurosurgeons surveyed. This is understandable given that transsphenoidal surgery is currently the standard surgical approach for the vast majority of sellar tumors, and anatomic abnormalities of the sphenoid sinus have the potential for serious intraoperative complications (e.g., injury to the internal carotid artery) [23‐26]. In this context, it should also be mentioned that a joint project of the American Society of Neuroradiology with the American College of Radiology and the Radiological Society of North America has created a set of common data elements for pituitary microadenomas, which similarly focuses on fundamental aspects that were identified as essential in our expert survey (including tumor location, size, contrast enhancement, and infundibulum abnormalities) [27, 28]. According to the experts interviewed, the tumor signal characteristics in the non-contrast T2 weighted imaging as well as T1 sequence after gadolinium application are significantly more relevant compared to the plain T1 weighted image. However, regarding the value of the different MRI sequences, the pertinent literature is quite heterogeneous. While Kumar et al. emphasize the importance of the non-enhanced T1 sequence and Karimian-Jazi the relevance of the dynamic T1 weighted imaging after contrast agent application for the detection of microadenomas, Bonneville considers the plain T2 sequence often more informative in pituitary imaging [6, 12, 29]. For tumor sizing, the clinical experts voted for the inclusion not only of the contrast-enhancing lesions, but additionally of any tumor cysts and necrotic areas present. This contrasts with the general practice for malignant gliomas, where, according to RANO criteria (Response Assessment in Neuro-Oncology), tumor cysts are generally not included in the measurement [30]. It should be noted here, though, that despite expansion of RANO efforts to include numerous neuro-oncology fields of work—such as low-grade gliomas, brain metastases, leptomeningeal metastases, meningiomas, and spinal neoplasms—no such recommendations have yet been developed for pituitary tumors [31]. Even beyond the RANO working groups, no standardized radiological criteria for the estimation of treatment response for pituitary tumors are available to date. However, in a recent study, Imber and colleagues were able to demonstrate adequate correlation of one- and two-dimensional measurement techniques with the volumetric gold standard, even for irregularly configured adenomas [32]. The utilization of the Hardy classification for a more detailed characterization of pituitary adenomas was rejected by most respondents. Originally developed in the 1970s using conventional radiographic techniques, significant limitations in terms of its reliability have been noted in the era of MRI despite frequent use in studies, so skepticism seems justified [33, 34]. Moreover, with a precise description of the tumor extension into the adjacent intracranial compartments and the skull base, all rated as essential by the participants, the key information contained in the Hardy classification is conveyed, so that a separate mention seems in principle dispensable. Thickening of the dura mater adjacent to the tumor (so-called dural tail) was considered an important element by neurosurgeons and radiotherapists in contrast to endocrinologists. This becomes understandable by the fact that besides an inflammatory origin, tumor cell infiltrates may well be present, which may have implications for local treatment modalities [35‐37]. Advanced imaging modalities such as MR perfusion and diffusion were also not considered integral parts of MRI protocols of sellar region tumors by the majority of participating experts. This is basically in line with contemporary recommendations from the related literature, in which performing thin-slice anatomic sequences (T1 weighted imaging with and without contrast and T2 weighted imaging) in multiple planes is considered a diagnostic imaging standard [3, 6, 7]. The Congress of Neurological Surgeons recommendations on imaging for nonfunctional pituitary adenomas emphasize the currently unclear importance of diffusion imaging regarding its correlation with tumor firmness, but also cite low-evidence findings that various MR perfusion techniques may provide information about adenoma vascularization, which could be valuable in terms of surgical planning and predicting the risk of postoperative bleeding [38]. Respondents had no clear preference on the question of the baseline to be selected for follow-up MRI. While imaging prior to initiation of therapy was chosen as a reference in a study on the response of invasive prolactinomas to bromocriptine, the current guideline of the AWMF (Association of the Scientific Medical Societies in Germany) advocates assessing tumor progression in comparison with the immediate past examination. Another option would be to choose the nadir of tumor extension as a comparator, as suggested by the RANO clinical trials working group in meningioma patients [39‐41].

The radiological report is a central communication tool in the interdisciplinary management of patients with space-occupying lesions of the sellar region. To meet the information needs of the involved clinical disciplines, MRI reporting should focus primarily on describing the location of the tumor epicenter, the size and exact compartmental extension, the associated impact on neighboring anatomical structures (especially the normal pituitary tissue/infundibulum, optic chiasm, cavernous sinus, internal carotid artery, and adjacent skull base), the internal characteristics of the mass, and the exclusion of complicating obstructive hydrocephalus. Apart from these basic requirements demanded by clinicians, additional characteristics may be relevant to the radiologist in making the correct differential diagnosis. Consistent adherence to a demand-oriented reporting checklist has the potential to improve interdisciplinary information exchange and thereby make a positive contribution to the care of this patient population.