To the best of our knowledge, this was the first database study to assess the prevalence of AGHD-related complications in GH-naïve patients with AGHD (diagnosis per claims database).

Women accounted for 56.5% of patients and showed two peaks in the histogram of age at diagnosis of AGHD, at ages 30 to 40 years and 60 to 70 years, whereas men showed only the latter peak. The bimodal distribution in women was found only in those without a history of pituitary tumor, and the women had a diagnosis of AGHD. Hypopituitarism can be easy to recognize in women because of menstrual abnormalities, whereas it can be more difficult in men, which could explain why no peak was seen in men at age 30 to 40 years. Another possible reason is that pregnancy-related hypopituitarism, including autoimmune hypophysitis and Sheehan’s syndrome, may be associated with this first peak.

About 17.4% of patients from the AGHD study cohort had malignant tumors. Although this implied a higher mortality risk in this subgroup compared to the general population, this reflected the actual conditions of AGHD patients from the real-world scenario and was in line with our primary goal.

In this study, we included patients with AGHD who were diagnosed just after being transferred from other hospitals and no data were available on previous pituitary surgery or radiation therapy. Thus, the proportion of patients with pituitary tumor, pituitary surgery, and radiation therapy may have been underestimated. Methodological differences between our and previous studies may explain some of the differences observed in these proportions [23].

The prevalence of AGHD-related metabolic complications in GH-naïve patients with AGHD was higher than that of metabolic complications in the general population. As AGHD causes metabolic disorder and an altered lipid profile [24], it is not surprising that dyslipidemia was the most frequent complication, with the greatest difference in prevalence between patients with AGHD and the general population (22.0% vs. 3.9%).

Estimated prevalences of complications were similar to or lower than those in previous studies. For instance, Mo et al. reported a prevalence of 14.0% for DM in GH-naïve patients with AGHD, which aligned closely with the prevalence in our study (9.3%) [25]. However, they found a higher prevalence of osteoporosis and depression than we did (10.7% vs. 4.8% and 16.1% vs. 3.4%, respectively), which could be due to a difference in study designs or advances in medical care. They also reported that 42.5% of untreated patients used lipid-lowering medication, which is higher than the prevalence of dyslipidemia (22.0%) in our study. The prevalence of these complications could also be underestimated in our study because of the nature of a database study. Incidences of myocardial infarction and angina in Japanese patients with AGHD were 1.2% and 2.8%, respectively, while those of cerebral infarction and cerebral hemorrhage were 3.7% and 0.8%, respectively [26], comparable to our results for ischemic heart disease (2.98%) and cerebrovascular disease (3.36%). Based on the patient demographics and complications observed in this study, we consider our findings to be representative of Japanese patients with AGHD.

Nonetheless, the prevalence of glucose intolerance and liver disease may have been especially underestimated because we could only identify complications recorded in the claims database. In 2020, the estimated prevalence of AGHD-related complications in patients with AGHD was higher than that in the general population from the Japanese national census. The prevalence of DM in our study (9.3%) was higher than that in the general population (4.2%) assessed from hospital-based data [27]. Furthermore, glucose intolerance was less common in patients with AGHD (4.1%) than in the general population with mean HbA1c value of 5.8% in men and 5.7% in females as reported per 2019 National Health and Nutrition Survey [18]. Our study found fewer cases of fatty liver (0.7%) and NASH/NAFLD (0.1%) than a previous study in the general population (27.7% for fatty liver, 23.7% for NAFLD) [28]. Liver disease is particularly prone to underreporting because the diagnostic procedures often require abdominal ultrasonography, computed tomography, magnetic resonance imaging, and liver biopsies, which are not always conducted in real-world clinical settings. Furthermore, because no established treatments are available for liver disease, medical costs may not be incurred by hospitals; hence, liver disease may not be recorded in the claims database. Terai et al. analyzed the MDV database and reported a prevalence of 1.93% for NASH/NAFLD in Japanese adults on a diagnosis basis [29]. On the other hand, Yoneda et al. estimated a much higher prevalence of 9.2% for NAFLD with the Fatty Liver Index prediction model by using medical check-up records in the JMDC database [30]. Therefore, the prevalence of liver disease may have been lower in this study because we could only identify complications recorded in the claims database, which may not have captured all cases.

We visualized AGHD-related complications with Kaplan-Meier curves. Although some previous studies focused on specific complications, to the best of our knowledge, ours was the first study to address a broad range of complications in patients with untreated AGHD. Unfortunately, in Japan, GHRT was contraindicated in patients with AGHD, not only those with malignant neoplasm but also in those with DM, up to 2022 including our study period. We prepared a Kaplan-Meier curve of AGHD both with and without these complications and found that the incidence of most complications was higher in patients with AGHD and DM or malignant neoplasm. However, the incidence of dyslipidemia was lower in patients with AGHD and malignant neoplasm than in those without. As dyslipidemia is one of the risk factors for malignancy, patients with malignant neoplasm could have a higher prevalence of dyslipidemia at index date than those without, which means fewer patients with malignancy would be newly diagnosed as dyslipidemia.

Although the effect of confounding variables cannot be completely ruled out, we explored the risk factors for complications using a multivariate Cox proportional hazard model adjusting for probable confounders. Although the prevalence of complications was higher in patients with AGHD than in the general population, the risk factors were similar in both groups. Age was a significant risk factor for most complications; being female, for osteoporosis and fracture; and DM, for dyslipidemia, ischemic heart disease, cerebrovascular disease, and all-cause death. The results of Cox regression captured the characteristics of each disease well, indicating that we used relevant definitions to identify each disease from the database.

Although it is crucial to acknowledge the complications associated with AGHD, it is also important to recognize the promising therapies available to mitigate these risks. GHRT is reported to ameliorate the risk of complications in patients with AGHD, e.g., studies reported reduced cardiovascular risk [31] and standardized mortality ratio [32] in patients with AGHD receiving GHRT. Improvement of body composition, lipid profile [33], and bone mineral density [34] were also described, and GHRT was found to prevent fractures [25]. Dyslipidemia, understandably the most frequent complication in our study, is not only one of the metabolic complications, but is also considered a significant contributor of cardiovascular risk among untreated GH-deficient adults today. Appropriate GHRT can positively impact the lipid profile [35, 36]. Furthermore, as assessed by the Adult Hypopituitarism Questionnaire, improved QoL was also reported in patients with AGHD [37].

In this study, we showed the real-world prevalence of AGHD-related complications in patients with untreated AGHD. It is noteworthy that only 8.1% of patients with AGHD received GHRT, a remarkably low percentage given the established benefit of GHRT. The low intervention rate suggests an unmet need for comprehensive treatment approaches in patients with AGHD; efforts are also needed to reduce the risk factors for related complications, to improve patient outcomes and quality of life. Taken together, the results of this database study and previous reports on each complication help us to provide more evidence-based guidance for the management of patients with AGHD. Avenues for further research include improved diagnostic criteria and patient management strategies for ensuring long-term benefits of GHRT.

This study has several limitations generally inherent to claims database research and specifically to the MDV database used to identify AGHD patients. Patients with AGHD and its associated complications were identified using ICD-10 codes, laboratory tests, and available medical records from the database. However, specific details essential for a definitive diagnosis of AGHD, such as GH secretion stimulation test data, IGF-I, medication details are often not captured in MDV, leading to potential misidentification of diseases. Indeed, this is a common limitation in database research [38], especially in therapeutic areas of rare diseases. MDV being a hospital-based claims database, patients cannot be traced if they change hospital because of relocation, and data on treatments not covered by insurance are not included. Further, some differences in the prevalence of certain complications with higher prevalences among the AGHD patients vs. those in the general population may be attributable to the way in which the data were collected: while the data on the AGHD patients were captured from the MDV database, the general population data were captured using the national census. Thus, the possibility of specialized follow-up and disease-specific testing conducted by endocrinologists managing AGHD leading to a higher detection of complications in the AGHD cohort cannot be denied. Nevertheless, we believe that our results provide the overall prevalence and impact of complications in a broader healthcare context by contrasting the data of an AGHD cohort vs. the general population.

Additionally, it should be noted that unlike rigorous comparative data from clinical trials, these real-world data are supportive in nature and may have limited generalizability. Despite these limitations, the study provides valuable insights into the prevalence and impact of complications. The precise identification of diseases and the establishment of comparison cohorts are critical considerations for future database research. Further discussions on the quality of databases and the refinement of analytical methodologies in database research will enhance its reliability.