We conducted a cross-sectional study recruiting 44 acromegalic adult patients referring to a single tertiary center. Patients were recruited between 2011 and 2023 and were diagnosed with acromegaly according to international guidelines (1, 15).

Thirty-one patients of our cohort were also included in a previous multicentric study, yet the present study includes 9 additional neuropsychological tests, exploring 5 additional cognitive domains. In particular, the following test have been added: Rey complex figure test, Rey Auditory verbal Learning test, Stroop test, Wisconsin Card Sorting Test, Weigl test, Semantic Fluencies, Naming of Object test, Token test, Attentional Matrices. Moreover, the inclusion of a healthy control group whose cognitive performance is directly compared to that of the patients’ group, enables us to study cognition not only from a quantitative approach, but also from a qualitative perspective.

We included acromegalic patients of age ≥18 years willing to perform a complete psychological and cognitive evaluation at recruitment. Exclusion criteria were: (i) a prior history of neurological or psychiatric diseases, (ii) assumption of psychiatric drugs, alcohol and/or drug abuse, (iii) severe visual or auditory impairments.

Moreover, we performed a prospective analysis of a small subgroup of 9 patients enrolled at diagnosis and followed-up, 6 of them up to 10 years. All patients underwent transsphenoidal pituitary adenoma surgery (TNS), which was performed at the Neurosurgery Unit of our Center. They underwent a complete psychological and cognitive assessment at four different time-points: (T1) at diagnosis, before any active intervention; (T2) 3–4 days post-operatively; (T3) 12-months after surgery; (T4) 10 years after surgery. In these cases, parallel versions of the tests were used when available, to avoid learning effect. A cohort of 40 adult healthy control subjects was also recruited, comparable for sex, age, and educational level. Controls were assessed with the same neuropsychological battery of patients. The previous presence of any medical or psychiatric condition which may influence the cognitive functioning was excluded by performing a complete psychological and medical anamnesis prior to the evaluation.

A written informed consent was signed by each participant. The local ethic committee approved the study (Comitato Etico Milano Area 2, approval number 973_2019bis, 10/10/2019).

Acromegalic patients completed the following self-report questionnaires investigating anxiety and depression:

In STAI X1 a score equal or superior to 65 or 71 was considered pathological for males and females, respectively. In the STAI X2 a cut-off of 56 and 62 represented pathological anxiety for males and females, respectively.

Both patients and controls completed an extended battery of 12 neuropsychological tests, addressing 14 cognitive functions, standardized for the Italian population, administered and corrected by a single dedicated neuropsychologist.

We tested short- and long-term verbal and visuo-spatial memory, selective attention and attentional shifting, verbal fluencies, inhibitory functions, constructional praxis, deductive reasoning and perseverative behavior. The complete test list with normality scores is detailed in Table 1. To avoid learning effect, parallel forms were used when available for the clinical practice.

The neuropsychological and psychological assessment took approximately 1 h; raw scores of the cognitive tests were corrected for age, educational level and gender when appropriate, according to Italian normative data, obtaining the adjusted scores (AS) used in the statistical analysis; equivalent scores (ES) were also calculated for qualitative analyses, as they enable to easily detect whether the performance is normal (ES = 2,3,4), borderline (ES = 1) or pathological (ES = 0), representing a ordinal scale. AS indicate the score obtained for each test and are represented by a continuous score.

AS, due to their test-specific scaling and psychometric properties, are difficult to compare directly across instruments and domains. In contrast, the use of a categorical parameter such as ES allowed us to schematize results across an extensive neuropsychological battery within a common metric, facilitating domain-level synthesis. For this reason, ES were adopted as the primary framework for cross-test and domain-level interpretation.

The cognitive evaluation administered to the prospective subgroup consisted in 4 additional tests investigating abstract reasoning: Weigl test (18), object-naming [Catricalà naming test (19)], verbal comprehension [Token test (20)] and visuo-spatial attention [Attentive Matrice (18)]. Such a wide battery was chosen in order to widely study the neuropsychological profile of patients and to overcome the limit of short assessments often reported in literature.

Clinical, demographical, neuroradiological and biochemical data were retrospectively extracted from medical records. IGF1 values were compared with an appropriate age- and sex-adjusted range and expressed as percentage of the upper limit of normal (ULN) for individual laboratory samples.

Acromegaly was considered in remission in the presence of normal IGF1 levels, measured at least 6 weeks postoperatively according to current guidelines, and GH values after OGTT <0.4 μg/L (1, 21, 22).

Disease status was recorded at diagnosis, at T3 (12 months after surgery) and T4 (10 years after surgery). Disease at T2 was not assessed as clinically and biochemically uninterpretable as to close to surgery as suggested form acromegaly guidelines (1).

Patients were considered controlled if during therapy mean fasting GH levels were below 1 μg/L in the presence of normal IGF1 levels (IGF1 < 1.2 × ULN), whereas in the remaining cases they were classified as uncontrolled (1, 22). IGF1 levels only were considered in patients receiving pegvisomant.

Patients were divided into three groups according to disease status at psychological evaluation as follows: active disease, comprising both naïve and uncontrolled patients, disease remission and controlled disease. As the main clinical outcome in acromegaly is the normalization of IGF-1 levels, for statistical purposes patients were grouped into two main categories: patients with elevated IGF-1 levels were considered as having active disease (including naïve patients and those uncontrolled during therapy), while patients with normal IGF-1 levels were considered as having controlled/cured disease (including both patients in disease remission and those controlled during therapy).

Hypopituitarism was defined as the presence of one or more hormonal deficit. Hypoadrenalism was defined by an abnormal response of cortisol in a dynamic test. In particular, before June 2016 we considered inadequate a peak of cortisol 500 nmol/L, during 1 μg corticotrophin stimulation test (ACTH 1 mcg) or insulin tolerance test (ITT); while after that time point the cut off according to the kit used in out Institution (Roche 2) changed to 351 nmol/L (23–25). A free thyroxine (fT4) under reference range in combination with not adequately increased TSH concentration was indicative of TSH deficiency (26, 27). Central hypogonadism was suggested in premenopausal women with low estradiol and low LH/FSH in conjunction with oligomenorrhea or amenorrhea, and in postmenopausal women in the presence of FSH levels inappropriately low for menopausal status. In men, central hypogonadism was defined by testosterone under reference range for age and normal/low LH/FSH in combination with clinical signs and symptoms (26). All hormonal deficits were appropriately treated with hormonal replacement therapy.

Continuous parameters were described as mean±standard deviation (SD) whe normally distributed and as median and interquartile range (IQR) when non-normally distributed. Normality of data distribution was assessed using the Shapiro–Wilk test. Continuous data were compared using t-test or Wilcoxon–Mann–Whitney test, as appropriate. Categorical data were presented as percentage (%), proportion (/) and analyzed using the Chi-squared test or Fisher’s exact test as appropriate.

Linear Regression analysis was performed to analyze the association between a series of endocrinological data (hormonal deficits, IGF-1 levels, disease status) and neuropsychological status, using the AS for quantitative analysis and the ES for qualitative descriptions. Friedman tests were conducted on longitudinal subgroup to examine the modification of cognitive function over time.

When Friedman test showed a significant overall effect, post-hoc Wilcoxon tests with Bonferroni’s correction were performed to account for multiple comparisons between different time-points.

p-values < 0.05 were considered statistically significant. Statistical analyses were conducted using GraphPad Prism 7.0 (GraphPad Software) and IBM SPSS 28.0 (IBM Analytics).

Demographical and clinical data of the patient’s cohort are detailed in Table 2.

The control group was composed by 40 healthy subjects (24F/16M), with a mean age of 52.43 ± 12.06 years and a mean educational level of 12.77 ± 3.11 years. These variables did not differ between patients and controls (p = 1.0, p = 0.089 and p = 0.494, respectively).

In the group of acromegalic patients, mean BDI score was not indicative of significant depressive symptoms (9.07 ± 8.08, range 0–39) considering its normative data. However, 10 patients (22.7%) reported a pathological score: in particular, 8 patients (18%) scored as mild and 2 patients (4%) as severe symptomatology. Fifty percent of patients scoring pathologically were in the active disease group (p = 0.13 vs. controlled/cured).

Analyzing results of anxiety questionnaires, the mean STAIX1 score in males was 53.17 ± 16.59, and in females was 50.42 ± 9.9, indicating a globally not significant state anxiety.

However, 6 males (33.5%) scored pathologically at STAIX1 test, while no female reported a clinical state of anxiety (STAIX1 female mean 50.42 ± 9.9). Four out of 6 patients (67%) with pathological scores were in the active disease group (p = 0.05 vs. controlled/cured), as shown in Figure 1A. On the contrary, mean results of STAIX2 indicated a slight general trait anxiety (52.22 ± 17.96), with 7 patients (38.9%) reporting a pathological score, 6 (85%) of whom were females. Even analyzing the STAIX2 results, the majority of patients with pathological scores were in the active group (70%) with a significant difference compared to the controlled/cured group (p = 0.017) (Figure 1B).

The linear regression analyses showed a significant correlation between IGF-1 levels and STAIX1 and X2 scores (p = 0.01, B = 0.64 and p = 0.008, B = 0.74 respectively), indicating that higher levels of the hormone are associated to higher levels of anxiety. On the contrary, no relation was found between the BDI test and IGF1 levels.

Interestingly, we also found a significant negative correlation between age and STAIX1 (p = 0.046, B = −0.29) and STAIX2 scores (p = 0.042, B = −0.33), indicating that younger acromegalic patients tend to experience more anxiety. No other clinical or demographical variables resulted significantly associated with depression or anxiety.

Results of the neuropsychological tests are reported in Table 3.

Overall, mean results of both patients and controls resulted within the respective ranges of normality in all the tests administered, considering AS; nevertheless, observing ES of the two groups, the number of patients scoring as pathological or borderline was significantly higher compared to controls in almost all tests, as shown in Table 3. This finding indicates that, besides obtaining a normal AS, patients perform poorer than healthy subjects in all the test considered, except for the Constructional Apraxia test.

Considering clinical data, hypopituitarism seems to impact the memory domain. In particular, observing ES, in the RAVLT-Immediate test 8 out of 15 (53%) patients with hypopituitarism scored as pathological versus only 4 patients out of 29 (14%) in the group without deficit (p = 0.010) and in the RAVLT-long term 5/15 patients (33%) with hypopituitarism scored pathological versus 1/29 patient (3%) without deficit (p = 0.01). In the Rey complex figure-recall test, we registered a slight difference in pathological results (ES) between hipopituitaric patients (scoring worse) and those without deficits (p = 0.05), with a stronger significance when including borderline results (p < 0.01).

Even if no differences were found between disease activity status and neurocognitive performances, IGF-1 levels still positively correlated with the results of Corsi test (AS) (p = 0.014, B = 0.04), suggesting an influence of hormonal values on the short-term visual memory functioning.

The prospective subgroup was composed by 9 patients: 4 (44.5%) males and 5 (55.5%) females, with a median age at diagnosis of 47 years (range 32–65) and a median educational level of 13 years (range 6–17). Not all patients completed the follow-up period, with 2 patients lost at T3 and 1 patient lost at T4 evaluation. At T3, 2/7 patients (28.5%) were in remission of disease and 2/7 (28.5%) were controlled with medical therapy.

At T4, 2/6 patients (33.3%) were in remission of disease and 4 (66.77%) were controlled with medical therapy.

Overall, patients scored within the respective normal ranges in most tests at baseline, and improved their neurocognitive performance over the follow-up period.

In particular, post hoc analysis with Wilcoxon signed-rank tests showed a significant improvement between T1 and T3 evaluations in the following tests: Rey-Osterrieth Complex Figure Test Recall [9.25 (1.25–18.6) vs. 21 (10.2–23.25), p = 0.01], RAVLT-Immediate recall [38.3 (27.2–44) vs. 53.7 (39.2–58.8), p = 0.03], RAVLT-long-term recall [7.7 (5.5–9.1) vs. 12.2 (9.6–13), p = 0.01], Semantic Fluencies [38 (35–45) vs. 49 (46–58), p = 0.03], TMT A [34 (29–42) vs. 30 (20–35), p = 0.01], TMT B [102 (84–172) vs. 85 (37–95), p = 0.03], Phonemic Fluencies [32 (26–43) vs. 47 (32–61), p = 0.01], WCST-perseverative score [19.5 (11.3–54.7) vs. 11.3 (10.4–23), p = 0.03], prose memory [8.5 (6.5–10.85) vs. 19.5 (8.5–21.25), p = 0.03].

Further significant effects emerged when comparing T1 and T4 for the TMT-A test [34 (29–42) vs. 23 (22–23), p = 0.01] and the WCST-Global score [66.5 (31.8–103.9) vs. 12.05 (11.2–23.4), p = 0.03]. As mentioned above, all these effects indicate a significant improvement over time, as shown in Figure 2.

Interestingly, in the Rey-Osterrieth Complex Figure Test Recall patients show a normal performance (mean 17.6 + 6.9) at T3 recovering from a borderline mean score at T1 (9.9 + 8.5).

No significant differences were found between T3 and T4 time points.

We acknowledge that this study is limited by the retrospective nature of the analysis. The small number of subjects in this study limit the conclusions that can be drawn and preclude definitive affirmations.