Patients
All patients included in the study were consecutive patients diagnosed with NFPA who underwent transsphenoidal microsurgery at one specialized center of pituitary surgery from October 2013 to January 2020 and who had pituitary MRI presurgery and three months after surgery available. All lesions were confirmed as NFPA by clinical and hormonal evaluation, and subtypes were defined by immunohistochemistry.
The cases were discussed in weekly multidisciplinary sessions with the presence of neuroradiologists, neuroendocrinologists, neuropathologists, radiotherapists, and neurosurgeons.
Endocrinological evaluation
All patients underwent investigation of hormone hypersecretion. IGF-I and prolactin (diluted 1:100 in adenomas ≥ 3 cm in the largest diameter to exclude hook effect) were measured for all patients, and during suspicion of Cushing's disease, tests were performed to investigate hypercortisolism, as recommended in the most recent guidelines18–20.
Additionally, basal cortisol was measured in the morning, and LH and FSH (in postmenopausal women and in men), testosterone (in men), TSH and free T4 were measured to evaluate the presence of hypopituitarism. Adrenal insufficiency was excluded in the presence of a basal cortisol level > 15.0 ng/mL and confirmed in the presence of values < 3.0 ng/mL. In patients with values between 3–15 ng/mL, replacement therapy was initiated, specifically if there were two other deficiencies or if central hypothyroidism was present (started in this case three days before levothyroxine replacement).
Surgery
Patients included in the study underwent a transsphenoidal approach, which was performed using the endoscope or with the microscope assisted with the endoscope. The surgeries were performed by three neurosurgeons, whom performed > 50 surgeries per year. Patients operated on by the subfrontal, transventricular or interhemispheric subcalous approach were not included in the study. The surgeons opted for these routes when the lesion had suprasellar extension compromising the Monro foramen or large lesions with extension to the anterior fossa and middle fossa.
During surgery, the objective was to safely remove as much of the lesion as possible, preserving the anatomical structures adjacent to the lesion and removing the mass effect generated by the tumor. The use of angled optics in the endoscope at the end of the surgery allows for a wide view of the sella region and identification of tumor residues to be removed.
When CSF leakage was visualized during surgery, abdominal fat was introduced into the surgical cavity after the adenoma was removed, using the nasal mucosa flap and the nasal bone septum, removed at the beginning of the approach, for reconstruction of the sellar floor. When CSF output was not observed during the procedure, the saddle floors were reconstructed only with the nasal septum bone.
Histopathology analysis
A neuropathologist evaluated all samples and reported histological features. Pathological examination included histological and immunohistochemical studies. The formaldehyde-fixed tissue was fixed in 10% buffered formalin, measured, and analyzed concerning its color and consistency, and then processed for paraffin embedding. Histological sections were stained with hematoxylin & eosin, and immunohistochemical analysis was performed by using antibodies for pituitary hormones and, when necessary, for transcription factors. In some selected cases, Gomori’s trichrome and/or reticulin stains were also performed. Antisera were directed against the following pituitary hormones (polyclonal): adrenocorticotropic hormone (ACTH; dilution 1:6000, cat. number 206A-76, Cell Marque, Rocklin, CA, USA), ß-follicle stimulating hormone (ß-FSH; dilution 1:6000, cat. number 207A-76, Cell Marque), growth hormone (GH; dilution 1:6000, cat. number 208A-76, Cell Marque), ß-luteinizing hormone (ß-LH; dilution 1:4000, cat. number 209A-16, Cell Marque), prolactin (PRL; dilution 1:7000, cat. number 210A-16, Cell Marque), ß-thyroid stimulating hormone (ß-TSH; dilution 1:4000, cat. number 211A-16, Cell Marque) and monoclonal α-subunit of glycoprotein hormones (α-SU; dilution 1:6000, cat. number ab11232, Abcam, San Francisco, CA, USA). When pituitary hormones and the α-subunit of glycoprotein hormones were negative (present in less than 1% of cells), immunohistochemistry of transcription factors was performed. The following monoclonal antibodies were used: SF-1 (dilution 1:400, cat. number PP-N1665-00, Perseus Proteomics, Tokyo, Japan), PIT-1 (dilution 1:2500, cat. number sc-393943, Santa Cruz Biotechnology, Santa Cruz, CA, USA) and T-PIT (dilution 1:900, cat. number AMAb91409, Atlas Antibodies, Stockholm, Sweden).
Radiological Evaluation
The patients underwent a preoperative MRI with contrast and a postoperative MRI with contrast 3 months after surgery. All preoperative and postoperative images were analyzed by a single specialized neuroradiologist and classified according to their invasion of the cavernous sinus, following the new Knosp graduation16, and the suprasellar, infrasellar, parasellar, anterior and posterior tumoral extension following the SIPAP classification17. The largest tumoral diameter was also measured in each MRI, and the largest tumoral diameter was ranked into 4 groups: 10–20 mm, 21–30 mm, 31–40 mm and > 40 mm.
All patients were scanned in the supine position using a dedicated 8-channel head coil for MRI acquisition in a 1.5T scanner (Magnetom Avanto, Siemens Healthcare and Optima 450w, GE Healthcare). MRI was performed using our sella imaging protocol, which included coronal and sagittal T1WI, T1 3D and T2 sequences (TR/TE 438/13 ms, FOV 20 cm, matrix 179 x 256, section thickness 2 mm and interslice GAP 0.1 mm) with and without gadolinium-based contrast agent (0.1 mmol/kg).
Patients were classified using postoperative MRI into two groups: patients with complete excision of the adenoma and patients with partial excision of the adenoma.
New combined score
A new combined radiological score was proposed based on the Knosp16 and the SIPAP classification17. We redesigned both schemes to carefully include tumor extensions into groups subjected to total tumor exeresis.
The classification of the extension of the pituitary adenoma was performed similarly to the classification of SIPAP17; however, we used the updated Knosp classification16 to classify the parasellar extension in “3A” and “3B”, as shown in Fig. 1, using the highest extension found on one side to classify the parasellar invasion, which in order of severity is 4 > 3B > 3A > 2 > 1 > 0.
Additionally, instead of the two groups of the SIPAP classification, we proposed a new division into three groups in this classification, which are divided as shown in Table 1.
Table 1
– New combined score for radiological prediction of surgical resection
| Knosp | Suprasellar | Infrasellar | Anterior | Posterior |
Group I | Grade 0 or 1 | Grade 0, 1 | Grade 0 or 1 | Grade 0 | Grade 0 |
Group II | Grade 2 and 3A | Grade 2 | Grade 2 | Grade 1 | Grade 1 |
Group III | Grade 3B and 4 | Grade 3,4 | Grade 2 | Grade 1 | Grade 1 |
The tumoral extension into the suprasselar, infrasselar, anterior and posterior directions are exemplified in Figs. 2, 3 and 4, respectively.
Group I included patients with grade 0–1 based on the Knosp classification, grade 0–1 in the suprasellar and infrassellar extension and grade 0 in the anterior and posterior extensions.
Group II consists of lesions that have Grade 2 and/or 3A in parasselar extension and/or Grade 2 in suprasellar extension and/or Grade 2 in infrasselar extension and/or Grade 1 in anterior or posterior extension.
Group III consists of lesions that present at least Knosp classification Grade 3B or 4 and/or suprasellar extension 3 or 4, as shown in Table 1.
Statistical analysis
Statistical analyses were performed using SPSS version 23.0 for Mac (IBM, Chicago, IL, USA). Categorical variables were expressed as percentages. For numerical variables, the median values (minimum–maximum) were used. The Mann‒Whitney test was used to compare numerical variables. The chi-square test or Fisher’s exact test was used to compare categorical variables, as appropriate. Binary logistic regression was performed for multivariate analysis. A p value < 0.05 was considered statistically significant.