In total, 383 patients were identified for this study; 256 (256/383; 67%) were men, and 127 (127/383; 33%) were women with a median follow-up duration of 8 years (IQR 5–10). The median age at presentation for the full cohort was 57 years (IQR 48–67).

The leading presenting manifestation was visual impairment (228/383; 60%). On formal ophthalmic assessment, 262 patients (262/383; 68%) had signs of visual field defect caused by optic compression. Hypopituitarism at presentation occurred in 58 patients (58/377, 15%); however, on preoperative endocrine screening (available for 377 patients), 235 patients (235/377; 62%) had evidence of at least one pituitary hormone deficiency. Sixty-six patients had their pituitary macroadenoma detected incidentally on radiological imaging (66/383; 17%) and due to headache in 41 patients (41/383; 11%). Twelve patients (12/377; 3%) were admitted with pituitary apoplexy.

With respect to radiological findings before surgery, data were available for 379 patients (Table 1). Fourteen patients (14/379; 4%) had a macroadenoma confined to the intrasellar space, 357 patients (357/379; 94%) had NFPMs abutting or compressing the optic apparatus, and 124 patients (124/379; 33%) had evidence of cavernous and sphenoid sinuses invasion (Table 1). NFPMs dimensions and volume on preoperative imaging were available for 317 patients; the correlation between tumour size with tumour recurrence and regrowth is shown in Table 2. Data of those with cavernous and sphenoid sinuses invasion were analysed separately and compared with the rest of the cohort (Table 3).

For primary surgery, most patients underwent transsphenoidal resection of NFPMs (378/383; 99%), and only five patients had transcranial surgery (5/383; 1%) (P < 0.001). Most of the patients in this cohort (360/383; 94%) underwent their primary surgery in our centre, while 23 patients (23/383; 6%) had their primary operation in other hospitals but were later referred to our hospital for ongoing treatment for recurrence or regrowth.

Following primary surgery, a total of 13 patients (13/383; 4%) had immediate surgical complications requiring intervention. These complications included eight patients with CSF leak (8/383; 2%), three patients with sellar haemorrhage (3/383; 1%), one patient with carotid injury (1/383; 0.3%), and one patient with an abdominal graft site haematoma (1/383; 0.3%). There were no postoperative deaths within 30 days of primary surgery in the cohort.

Total resection was achieved in 229 patients (229/383; 60%). The rate of complete resection improved over time; in the last quartile of cases, 77 patients achieved complete resection (77/95; 81%). Of those that achieved complete resection, 28 patients (28/229; 11%) had tumour recurrence on imaging after a median follow-up of 3 years (IQR 2–5). Six of these patients (6/229; 3%) had tumour recurrence requiring a second surgery after a median duration of 7 years (IQR 6–10) after their first surgery. Re-operation was via the transsphenoidal approach in all cases. Surgical reintervention-free survival probability for those who had complete resection at 5, 10 and 15 years was 99%, 94% and 94%, respectively (Figure 2).

Residual disease was evident in 154 patients (154/383; 40%) on the first postoperative imaging; of those, 106 (106/154; 69%) had tumour regrowth on imaging after a median follow-up of 3 years (IQR 1–5) and 84 patients (84/154; 55%) needed a second surgery after a median duration of 6 years (IQR 2–9) of their first surgery. Transsphenoidal resection for adenoma recurrence or regrowth was performed in 80 patients (80/84; 95%), while only four patients had transcranial operation (4/84; 5%). Surgical reintervention-free survival probability for patients who underwent incomplete resection of NFPM at 5, 10, and 15 years was 74%, 49% and 35%, respectively (Figure 2).

Patients with residual disease following primary surgery had a higher risk of tumour regrowth than those who underwent complete resection with an Odds ratio of 16 (95% CI = 10–28) (P = <0.001). Younger patients were more likely to undergo secondary surgery with a median age of 53 years (IQR 44–61) vs. those patients who didn’t need surgical reintervention at 59 years (IQR 49–68) (P = <0.001). Cavernous and sphenoid sinuses invasion on the first postoperative imaging didn't influence the need for further surgical intervention in those with and in those without features of invasion (0.2).

For the entire cohort, larger tumours, as demonstrated on pre-operative imaging, were associated with an increased risk of recurrence and regrowth (Table 2). Younger patients had a higher risk of disease recurrence (P ≤ 0.001). Patient's sex and choice of micro-or endoscopic surgical approach were not shown to be risk factors for tumour recurrence and regrowth (Table 2).

Based on hormone expression using immunohistological analysis of the resected tumours, Gonadotroph-expressing NFPMs occurred in 371 patients, while 12 patients had pleurihormonal pituitary adenomas. The extent of Ki-67 expression in the resected pituitary adenomas is shown in Table 4. It was higher in those who required a secondary surgery, but this was not statistically significant; 23% (77/335) of patients with Ki-67 < 3% required re-operation, compared to 30% (13/43) of patients with Ki-67% > 3% (P = 0.3). Nor was Ki-67 associated with the timing of the second surgery (P = 0.2). However, Ki-67 was higher in those that received postoperative radiotherapy than in those treated with surgery alone (P ≤ 0.001) (Table 4).

There was a significant inverse relationship between high Ki-67 and age (P = 0.01) (Table 4). Ki-67 expression was similar in those with preoperative visual loss and those with normal vision (P = 0.1). In addition, high ki-67 more than 3% was not associated with cavernous sinus invasion on preoperative imaging (13/111; 14% with cavernous sinus involvement vs. 30/263; 11% in those without) (P = 0.9).

Pituitary radiotherapy to control tumour regrowth was delivered to 65 patients (65/383; 17%) after a median time of 1 year (IQR 1–3) following surgery. Radiotherapy was used more to control tumour progression in those who had regrowth of their residual disease (61/106; 58%) than those patients who had NFPMs recurrence after complete resection (4/28; 14%) (P ≤ 0.001). External beam radiotherapy was administered in 63 patients (50.4 Gray in 28 daily fractions over 5.5 weeks), and single-fraction Gamma knife radiosurgery was performed in two patients. One patient (1/65; 2%) showed evidence of tumour regrowth after 23 years from receiving radiotherapy, and seven patients (7/65; 11%) had NFPMs regression on imaging after radiotherapy. None of those treated with irradiation needed further surgery. Four patients (4/65) had a complete resection.

Tumour progression-free survival probability after irradiation was 100% at 5, 10 and 15 years (Figure 3). Postoperative radiotherapy was used less in older patients to control residual tumour (P = 0.003) and was delivered more often in patients with cavernous and sphenoid sinuses invasion (39/124; 32%) on postoperative imaging than those without (26/255; 10%) (<0.001).

Normal visual assessment was reported in 110 patients (110/383; 29%), while 263 patients (263/383; 69%) had evidence of visual impairment before surgery. Nine patients (9/383; 2%) with underlying eye disease were excluded. At the latest review, 81 patients with preoperative visual deficit (81/263; 31%) had normal vision, 155 patients (155/263; 59%) had visual improvement, 18 patients (18/263; 7%) retained their visual impairment, and one patient (1/263; 0.4%) suffered permanent deterioration following surgery. There were no reported cases of radiotherapy-induced optic neuropathy.

This retrospective cohort study of 383 patients with NFPMs reports the following principal findings: (1) Residual disease on the first post-operative scan was a significant risk factor for NFPMs regrowth and associated with a higher risk of re-operation and radiotherapy; and (2) Ki-67 antigen expression in the resected pituitary adenoma was not a significant risk factor for re-operation, albeit higher in young patients and those that went on to receive radiotherapy. With these findings in mind, we propose a follow-up strategy that stratifies patients at “low risk” if there is no residual adenoma, with increasing scan intervals, or at “high risk” if there is a residual disease, with annual scans for at least five years and extended surveillance after that (Figure 4). In patients with growing residual disease, we consider re-operation if there is compression of the optic apparatus; and re-operation and/or radiotherapy in patients who are young enough such that future compression is anticipated.

Pituitary adenomas are re-termed PitNET in the 5th Edition of the WHO Classification of Endocrine and Neuroendocrine Tumors (1). Non-functioning subtype is shown to recur following surgery more frequently than other types of pituitary adenomas (11). The lack of hormone hypersecretion-related syndromes impedes early diagnosis and leads to delayed presentation with large tumours. NFPMs commonly manifest by invading nearby critical structures and compressing the optic pathway. Total resection of NFPMs is challenging, particularly while operating on invasive tumours extending beyond the sella turcica. Therefore, long-term follow-up with regular clinical and radiological assessment is necessary to detect disease progression and manage with further surgery with and without radiotherapy when appropriate.

Several studies attempted to establish robust clinicopathological and radiological factors to predict NFPMs recurrence after surgery. However, there is no consensus about these variables (16). Consequently, defining a clear neuroimaging time frame after surgery when following patients with NFPMs is still lacking. Young age, cavernous sinus invasion and partial resection of NFPMs are independent predictive factors of recurrent growth detected in this study. Similar findings were demonstrated by others (17–19).

On the contrary, Yeo et al. (20) and Batista et al. (21) studies have not related age to tumour recurrence, whereas Dubois et al. (22) reported that remanent tumours didn't influence recurrence in non-functional gonadotroph adenoma. In general, younger patients are more likely to have closer surveillance and radical treatment when developing NFPMs progression following surgery, while elderly patients have a more conservative approach. Patients with residual disease were at higher risk of further surgical intervention for regrowing tumours than those who underwent total resection. In addition, as demonstrated in this study, those with remanent tumours required surgical resection earlier than residual-free patients.

Transsphenoidal surgery is used in resecting most pituitary tumours. To date, two operative visualisation techniques exist: microscopic and endoscopic. The operating microscope was introduced in the late 1960s (23) and was the gold standard for resecting pituitary tumours over the next 30 years. Endoscopic transsphenoidal surgery is an alternative route introduced in the 1990s (24) with two approaches, endonasal and sublabial. There has been a growing trend of resecting pituitary adenomas using the endoscope over the last two decades during primary surgery and resecting progressing residual disease, given its enhanced visualisation, particularly into the cavernous sinus and suprasellar region, offering the potential for a greater degree of tumour removal than the microscopic approach. To this end, many studies have demonstrated a higher rate of complete surgical removal of pituitary adenomas, the most crucial prognostic outcome, using endoscopic surgery than microscopic modality (25–27). In this cohort, early patients predominantly underwent microscopic pituitary surgery, as two of the neurosurgeons operated using this approach. Endoscopic surgery was introduced in our centre in 2012, and all the current neurosurgeons are using this technique, with a corresponding improvement in resection rates demonstrated in this cohort. Transcranial surgery is rarely indicated but may have a role for patients with large pituitary adenomas and substantial temporal or anterior cranial fossa extension (28). Combined transsphenoidal and transcranial approaches can also be used in some circumstances.

Computer-guided neuronavigation and intraoperative imaging have also been increasingly deployed in pituitary surgery to achieve the best surgical outcome. Conducting transsphenoidal surgery with intraoperative MRI remains contentious due to increased cost and length of surgery, but it may more readily demonstrate the extent of tumour resection and increase the rate of complete resection, particularly in large tumours and in cases of recurrent adenomas (29, 30). In addition, MRI-guided resection is shown to have a low recurrence rate (29). Other intraoperative imaging using CT or ultrasound guidance have also been used, with the latter offering the advantage of being more readily integrated into the operative workflow and avoiding ionising radiation (31).

Definite molecular risk factors for NFPMs recurrence still need to be well established. The ki-67 antigen is considered a marker of tumour proliferation and invasiveness in pituitary tumours (32). However, its prognostic value in predicting recurrence and regrowth is debatable (32–34). The World Health Organization's classification of pituitary adenomas and recommendation for the behaviour of invasive tumours propose a cut-off of Ki-67 proliferation index >3% (35) as a factor of aggressive pituitary adenomas. In this study, recurrent NFPMs had a higher Ki-67 (>3%) than non-recurrent tumours, albeit not significantly raised. Furthermore, the proliferation index was higher in younger patients and those that went on to receive radiotherapy but not in those with cavernous and sphenoid sinus invasion compared to patients with macroadenoma confined to the sellar space. The relationship between Ki-67 antigen and age, pituitary adenoma recurrence and aggressive tumour behaviour has been controversial. Kyung-II Paek et al. (34) and Mastronardi et al. (36) couldn't correlate between Ki-67 and age in patients with pituitary adenomas. In contrast, other studies recorded higher expression in young patients (22, 37, 38), which is concordant with our study. Hongegger et al. (39) reported an independent correlation between Ki67 and invasiveness in NFPAs, while Chiloiro et al. (40) concluded that Ki-65 expression of more than 1.5% is associated with a greater risk of cavernous and sphenoid sinus invasion, which is in contrast to our findings, but their study included a cohort of functioning and non-functioning pituitary adenomas which can bias their data. Trouillas et al. (33), in their multicentric case-control study, developed the five-tiered prognostic clinicopathological classification of pituitary adenomas. The classification is based on pituitary MRI for cavernous or sphenoid sinus invasion and tumour size, immunocytochemistry analysis and proliferation markers (Ki-67, mitoses, and P 53). Tumours were classified according to size (microadenoma, macroadenoma and giant pituitary adenomas), type (Prolactin, Growth Hormone, FSH/LH, ACTH and TSH) and grade (grade 1a: non-invasive, 1b: non-invasive and proliferative, 2a: invasive, 2b: invasive and proliferative, and 3: metastatic). Sahakian and Raverot et al. (1, 21) assessed and validated the impact of this classification on tumour progression in invasive and proliferative PitNETs.

Postoperative pituitary radiotherapy provides excellent control of non-functioning adenoma regrowth with a potential reduction in tumour volume in a small proportion of cases. However, its routine use has been controversial and limited as adjuvant therapy for progressive and invasive tumours when surgery is unfeasible. In this study, the probability of tumour progression-free survival after radiotherapy was comparable to the literature (8). Age and cavernous sinus involvement influenced the need for radiotherapy for our patients. Chang et al. (41) also reported that those with adenomas invading the cavernous sinus needed radiotherapy more frequently than those without (41). The potential risk of hypopituitarism, injury to nearby radiosensitive structures, late development of secondary brain tumours and cardiovascular accidents should be considered when radiotherapy is used. In our centre, all patients are discussed in the pituitary multidisciplinary meeting to assess the need for surgery with careful consideration of using radiotherapy. We currently deploy proton beam therapy to treat children and young adults up to their 25th birthday to reduce the late effects of radiation.

Surgery is very effective in relieving compressive visual symptoms, with 90% of patients having normal or improved symptoms after presenting with a visual deficit with an extremely low incidence of neuro-ophthalmic deterioration, as demonstrated in this series. This is keeping with the literature (42–44). Radiation-induced optic neuropathy is a rare complication of pituitary adenoma irradiation (8, 45); none were reported in this cohort.