A retrospective cohort investigation was carried out on consecutive patients with suprasellar GCTs admitted to Beijing Tiantan Hospital, Capital Medical University, between September 2015 and August 2023, who required an endocrine follow-up period of at least 3 months. Patients who lacked important clinical data, were receiving oncologic treatment, or had completed less than 3 months of oncologic treatment were excluded. During this period, a total of 567 patients with suprasellar GCTs were identified. Among them, 48 patients were excluded due to a lack of essential clinical data. This included 13 cases with unclear medical history and treatment information, 17 cases lacking initial imaging data, and 18 cases without any neuroendocrine function assessment data. Additionally, 26 patients undergoing oncologic treatment and 34 patients receiving treatment for less than 3 months were excluded because they could not accurately assess post-treatment neuroendocrine function. Finally, 459 patients with suprasellar GCTs were enrolled, as depicted in the screening flowchart presented in Figure 1 . In prior research on iGCTs, delayed diagnosis was considered to be the period from the first presentation to the definitive diagnosis lasting more than 6 months (11, 15, 16, 22). Therefore, this study divided all individuals into two groups based on the duration from the onset of the first symptom to definitive diagnosis: delayed diagnosis (>6 months) and non-delayed diagnosis (≤6 months). Data were collected retrospectively, including sex, age at onset, clinical symptoms (endocrinological symptoms, neurological symptoms, and symptoms associated with mass effects), tumor characteristics (maximum tumor diameter, tumor location, type, and metastasis), diagnosis, and treatment. Neuroendocrine function assessments were recorded at diagnosis, at the end of oncologic treatment, and at the last follow-up.

In this study, suprasellar GCTs were diagnosed based on histopathological findings in 189 patients (41.2%), elevation of serum/cerebrospinal beta-human chorionic gonadotropin (β-hCG) and/or alpha-fetoprotein (AFP) levels in 223 patients (48.6%), and diagnostic radiotherapy (typical clinical characteristics and imaging findings, along with a reduction of exceeding 80% in the maximum lesion diameter following low-dose local radiation therapy) in 47 patients (10.2%). Patients who had β-hCG levels above 100 IU/L or elevated AFP were deemed to be NGGCTs if no histopathological findings were present, whereas the rest of the patients were considered to be germinomas.

After diagnosis, patients underwent 2 cycles of platinum-based induction chemotherapy (ifosfamide, 1.5 g/m2, d1–3; etoposide, 70 mg/m2; d1–3; and cisplatin, 30 mg/m2; d1–3) administered every 4 weeks. Subsequently, radiotherapy was administered, followed by an additional 2 to 4 cycles of chemotherapy. Regarding the radiation field and dose in our institute for germinoma, a whole-brain/whole-ventricle radiotherapy (24.0–30.6 Gy) was followed by a boost to the tumor region for a total dose of 40 Gy. For NGGCTs, whole-brain/whole-ventricle radiotherapy (30.6 Gy) was followed by a boost to the tumor region for a total dose of 54 Gy. Craniospinal irradiation plus boost therapy was reserved for patients with metastatic disease. In clinical practice, the specific irradiation dose and field vary depending on patient age, tumor size, tumor extent, and serum levels of β-hCG and AFP. Overall, 206 individuals (44.9%) underwent surgery, of which 109 (23.7%) underwent stereotactic biopsy, 66 (14.4%) underwent surgical resection of lesions, 13 (2.9%) underwent biopsy + endoscopic third ventriculostomy (ETV) or ventriculoperitoneal shunt (VPS), and 18 (3.9%) underwent ETV or VPS. Among the 459 patients included, 428 (93.2%) received chemotherapy and all patients underwent radiotherapy. The irradiation dose and field were recorded in 291 patients, with a median total dose of 40 Gy (range:30.4 Gy-56 Gy); 187 patients (64.3%) received whole-brain/whole-ventricle radiotherapy, while 104 (35.7%) received craniospinal irradiation.

Neuroendocrine functions such as the hypothalamus-pituitary-adrenal (HPA), hypothalamus-pituitary-thyroid (HPT), hypothalamus-pituitary-gonadal (HPG), growth hormone (GH)/insulin-like growth factor I (IGF-1), hypothalamus-pituitary-prolactin axis and neurohypophysis were evaluated. Thyroid hormone and thyroid-stimulating hormone were measured using Beckman chemiluminescent immunoassay (Brea, CA, USA). Serum IGF-1, GH, adrenocorticotropic hormone, cortisol, estradiol, testosterone, luteinizing hormone (LH), follicle-stimulating hormone (FSH), and prolactin were determined using chemiluminescent immunoassays (2000xpi system, SIEMENS). Cortisol serum levels below 3 µg/dL at 8:00 am or peak cortisol levels of less than 18 µg/dL during an insulin-induced hypoglycemia test were considered diagnostic of central adrenal insufficiency (CAI) (23). Central hypothyroidism (CHT) was defined as low free thyroxine and low or normal thyrotropin levels. Central hypogonadism was diagnosed in females over 13 years old and males over 14 years old when serum estradiol or testosterone levels were decreased, accompanied by low levels of LH and FSH (23). However, HPG axis function was not evaluated in patients without puberty initiation in children (males <9 years old; females <8 years old) or those in early pubertal stages (9–14 years in males; 8–13 years in females) (24). Growth hormone deficiency (GHD) was defined as a serum IGF-1 level lower than the normal reference range for age and sex, accompanied by a concomitant lack of three other pituitary hormones, or peak GH levels <5.0 μg/L in the insulin hypoglycemia stimulation test (23). After excluding stress and medication factors, prolactin serum levels exceeding the upper limit of normal reference values were considered hyperprolactinemia (HPL). Arginine vasopressin deficiency (AVP-D) was diagnosed when plasma osmolality was >295 mOsm/kg and corresponding urine osmolality was <300 mOsm/kg in fluid deprivation tests, and the symptoms of polyuria/polydipsia were alleviated after administration of vasopressin (23, 25). Precocious puberty refers to the early development of secondary sexual characteristics before the age of 8 in girls and 9 in boys (24, 26). Delayed puberty was characterized by a lack of puberty initiation until age 13 for girls and 14 for boys (27). The timing of neuroendocrine function assessment at diagnosis occurred prior to any treatment, including radiotherapy, chemotherapy, and surgery. The neuroendocrine assessment at the end of oncologic treatment was conducted within one month after completing tumor treatment. Routine follow-up assessments were conducted every 3–6 months during the initial 2 years following oncologic treatment, and thereafter, every 6–12 months. Patients were followed-up until December 2023.

Recognizing the significant influence of impaired neuroendocrine function on patient survival and quality of life, we implemented a four-grade classification system to assess the severity of neuroendocrine dysfunction. We also delineated the most prevalent types of pituitary dysfunction in different grades to guide the clinical practice. Grade 0 indicated normal neuroendocrine function with no impairment of the adenohypophysis or neurohypophysis function. Grade 1 indicated mild impairment in the presence of one or more components of GHD, CHG, HPL, and AVP-D. The primary type of pituitary dysfunction in this grade was AVP-D with or without one or more components of GHD, CHG, or HPL (denoted as AVP-D +/- GHD/CHG/HPL). Grade 2 signified moderate impairment in the presence of CHT, excluding CAI, with or without one or more components of GHD, CHG, HPL, or AVP-D. The most common pituitary dysfunction in grade 2 was the combination of CHT with AVP-D with or without one or more components of GHD, CHG, and HPL (indicated as CHT+AVP-D+/-GHD/CHG/HPL). Grade 3 indicated severe impairment in the presence of CAI with or without one or more components of CHT, GHD, CHG, HPL, and AVP-D. Within grade 3, panhypopituitarism (CAI+CHT+ AVP-D +GHD+CHG) was the primary type of pituitary dysfunction.

For continuous variables that were not normally distributed, the median and interquartile range were reported, and the Mann-Whitney U test was used to perform comparisons. Categorical variables are presented as numerical values and percentages, and comparisons were performed using chi-squared or Fisher’s exact tests. To explore the independent effect of delayed diagnosis on the grade of neuroendocrine dysfunction at various stages, we conducted univariate and multivariate ordinal logistic regression analyses. In this analysis, the grade of neuroendocrine dysfunction was considered the dependent variable, while delayed/non-delayed diagnosis was treated as the independent variable. Our analyses meticulously accounted for potential confounding variables, like age of onset, sex, tumor type, tumor location, and maximum tumor diameter at diagnosis. We additionally adjusted for surgery at the end of oncologic treatment. Furthermore, the irradiation dose and those potential confounding factors mentioned above were adjusted at the last follow-up. Given that only 291 patients had detailed records of radiotherapy doses in this study, the multivariate ordinal logistic regression analyses at the last follow-up included only these 291 patients. A 2-tailed P value <0.05 was considered statistically significant. SPSS 25.0 (SPSS Inc.) was used for all statistical analyses.

Based on the duration from initial symptoms to definitive diagnosis, 325 patients (70.8%) were assigned to the group with delayed diagnosis, while 134 (29.2%) were in the group with non-delayed diagnosis. Of the 459 enrolled patients, 244 (53.2%) were female and 106 (23.1%) exhibited “bifocal” lesions. The median age at onset and maximum tumor diameter were 11 (8–15) years and 20 (12–32) mm, respectively. A total of 350 patients (76.3%) were diagnosed with germinoma and 109 (23.7%) were diagnosed with NGGCTs. Tumor dissemination was observed in 79 patients (17.2%). The median endocrine follow-up duration was 24 months. There were 183 patients (40.0%) with follow-up durations exceeding 36 months, 123 cases (26.8%) exceeding 48 months, 92 cases (20%) exceeding 60 months, and 61 cases (13.3%) exceeding 72 months. During the follow-up, 19 patients (4.1%) experienced a relapse, among which 2 cases died due to disease progression. Among the 17 surviving patients, 14 underwent additional radiotherapy and chemotherapy, while 3 patients received a combination of chemoradiotherapy along with surgical resection of residual lesions. Individuals with delayed diagnosis had more females, more isolated suprasellar lesions, a larger maximum tumor diameter, and a greater prevalence of germinomas than those with non-delayed diagnosis ( Table 1 ).

In terms of endocrinological symptoms, polyuria/polydipsia was reported in 407 cases (88.7%), slow growth in 109 (23.3%), fatigue in 99 (21.6%), amenorrhea in 69 (15%), delayed puberty in 43 (9.4%), and precocious puberty in 17 (3.7%). Regarding the clinical manifestation indicative of mass effects, visual acuity/field changes were noted in 145 cases (31.6%), headaches in 117 (25.5%), and nausea or vomiting in 70 (15.3%). However, neurological symptoms were relatively infrequent. As for the duration of symptoms, all endocrinological symptoms exhibited relatively prolonged durations. Slow growth persisted for a substantial 24-month period. Subsequently, delayed puberty, amenorrhea, and polyuria persisted for 12 months each, with fatigue being the shortest at 3 months. However, the duration of mass effects and neurological symptoms was significantly shorter than that of the endocrinological symptoms ( Figure 2 ). Moreover, individuals with delayed diagnosis showed a greater frequency of endocrinological symptoms, including slow growth, polyuria/polydipsia, fatigue, and amenorrhea, when compared to those with non-delayed diagnosis. Conversely, they experienced a lower incidence of symptoms such as precocious puberty, headaches, and nausea/vomiting ( Table 1 ).

A comparison of neuroendocrine dysfunction between the two groups at different stages is shown in Table 2 . Upon diagnosis, AVP-D had the highest prevalence (96.4%), followed by CHG (83.2%), HPL (71.6%), GHD (68.8%), CAI (64.8%), and CHT (61.7%). After the end of the oncological treatment, the incidence of HPL decreased significantly to 24.5% at the last follow-up. However, the prevalence of AVP-D, CHG, GHD, CAI, and CHT remained high at the last follow-up, with rates of 93.8%, 65.3%, 71.4%, 60.8%, and 68.9%, respectively. More CAI, CHT, AVP-D, GHD, CHG, and HPL were observed in the delayed diagnosis group at diagnosis, at the end of oncologic treatment, and the last follow-up compared to those in the non-delayed diagnosis group, except for HPL at the last follow-up.

Moreover, considering the large variation in the follow-up period of patients and the time-dependent effects of radiotherapy on neuroendocrine function, we further compared the neuroendocrine function between the two groups at different follow-up periods. The results revealed that regardless of whether it was at the 24-month, 48-month, or 72-month follow-up, patients with delayed diagnosis still exhibited a higher incidence of CAI, CHT, AVP-D, GHD, and CHG than those with non-delayed diagnosis ( Table 3 ).

A comparison of grades of neuroendocrine dysfunction and their corresponding primary types of pituitary dysfunction in the two groups at various stages is presented in Table 4 . At diagnosis, the most common grade of neuroendocrine dysfunction was grade 3 (285 cases, 64.4%). This was followed by grades 1, 2, and 0 in the order of 123 (27.7%), 32 (7.2%), and 4 (0.9%) cases, respectively. After oncologic treatment and during the last follow-up, the prevalence of grade 3 slightly decreased, yet it remained the predominant category. Except for the lack of statistical significance in the difference in grade 0 at diagnosis or grade 2 at the last follow-up between the two groups, individuals with delayed diagnosis were more likely to have grade 3 neuroendocrine dysfunction and less likely to have grade 0–2 neuroendocrine dysfunction when compared to those with a non-delayed diagnosis at diagnosis, end of oncologic treatment, and last follow-up. Among the primary types of pituitary dysfunction among grades 1–3, aside from the absence of statistically significant disparities in CHT+ AVP-D +/- GHD/CHG/HPL between the two groups at the last follow-up, individuals with delayed diagnoses had more panhypopituitarism and a lower prevalence of AVP-D +/- GHD/CHG/HPL and CHT+ AVP-D +/- GHD/CHG/HPL, whether at the diagnosis, end of oncologic treatment, or last follow-up.

To investigate the independent effect of delayed diagnosis on the grades of neuroendocrine dysfunction, we conducted an ordered logistic regression analysis. Univariate analysis revealed that individuals with delayed diagnosis exhibited higher grades of neuroendocrine dysfunction at diagnosis, end of oncologic treatment, and last follow-up. Following adjustments for other potential confounding factors, this conclusion remained evident, both at the time of diagnosis (OR=3.005, 95% CI 1.929–4.845, p<0.001), at the end of oncologic treatment (OR=4.802, 95% CI 2.878–8.004, p<0.001), and during the last follow-up (OR=2.335, 95% CI 1.307–4.170, p=0.005) ( Table 5 ).

We further explored the recovery of the impaired neuroendocrine function after treatment. The results showed that, compared to the time of diagnosis, a certain proportion of patients recovered from neuroendocrine dysfunction during follow-up. Notably, HPL exhibited the highest recovery rate of 67.3%, followed by CHG at 22.6%, CAI at 18.3%, CHT at 10.0%, and GHD at 6.1%. AVP-D displayed the lowest recovery rate of 3.7% ( Table 6 ). Taking the end of oncologic treatment as the starting point, the recovery times for AVP-D, CHG, and GHD were relatively long, with a median of 12 months (range: 0–27 months), 10 months (range: -4 to 48 months), and 9 months (range: 0–36 months), respectively. In contrast, CAI and CHT had relatively short recovery times, with a median of 1 month (range: -4 to 24 months) and 0 months (range: -4 to 48 months), respectively ( Figure 3 ). Individuals with a delayed diagnosis experienced significantly lower recovery rates for AVP-D, CAI, and CHT than those with a non-delayed diagnosis. Although there were also lower recovery percentages in CHG, GHD, and HPL, these differences did not reach statistical significance ( Table 6 ).

During the period from diagnosis to the end of oncologic treatment, 64 patients (13.9%) developed new-onset neuroendocrine dysfunction. The incidence of newly developed AVP-D, CAI, CHT, CHG, and GHD was 4/11 (36.3%), 24/138 (17.4%), 32/154 (20.8%), 2/35 (5.7%), and 18/109 (16.5%), respectively. A higher proportion of surgical resection or biopsy was observed in newly developed cases than in those without newly developed dysfunction (53.1% vs 39.0%, p=0.033). During the follow-up period from oncologic treatment completion to the last follow-up, 55 patients (12.0%) experienced new-onset neuroendocrine dysfunction. Among them, at the follow-up time points of 24, 36, 48, 60, 72, and 84 months, there were 12 cases (2.6%), 31 cases (6.8%), 38 cases (8.3%), 43 cases (9.4%), 48 cases (10.5%), and 55 cases (12.0%), respectively. The incidences of newly developed AVP-D, CAI, CHT, CHG, and GHD were 6/27 (22.2%), 23/151 (15.2%), 24/142 (16.9%), 6/60 (10%), and 20/113 (17.7%), respectively. The median radiation dose was higher in the newly developed dysfunction group than in the non-newly developed dysfunction group (44.6Gy vs 40.0 Gy, p = 0.037).