Relationship between COVID-19 and hypopituitarism: A two-sample Mendelian randomization study

Abstract

Objectives

COVID-19 has been identified as causing damage to the nervous and endocrine systems, with observed effects on hypopituitarism. However, the effects of COVID-19 on hypopituitarism are controversial. In this study, we conducted MR analysis to investigate the relationship between COVID-19 and hypopituitarism.

Study design

The MR study utilized publicly available summary statistics from an independent European cohort. Data were sourced from the EBI database for COVID-19(case N = 8316 and control N = 1549,095) and from the FinnGen database for hypopituitarism (case N = 910 and control N = 211,123). Independent genetic variants significantly associated with each exposure(P < 5 ×10–8) were considered as instrumental variables. The MR analysis and colocalisation analysis were underwent.

Methods

Four methods were employed: IVW, MR-Egger regression, weighted median and weighted mode with sensitivity analyses conducted. Finally, colocalisation analysis was also conducted.

Results

MR study analyses were performed using the IVW fixation method, and we found a significant causal association between COVID-19 and hypopituitarism. IVW OR= 1.283, 95 %CI(1.017, 1.617), p = 0.035; MR-Egger OR= 1.355, 95 % CI(0.850, 2.160), p = 0.291; OR= 1.327, 95 %CI(1.017, 1.732), p = 0.037 in the weighted median approach and OR= 1.330, 95 % CI(0.993, 1.782), p = 0.128 in the weighted modal approach, with the results of the sensitivity analysis showing a stable and suggestive of a causality that COVID-19 increases the risk of developing hypopituitarism. The probability of colocalization of COVID-19 and hypopituitarism was PP.H4.abf= 9.55 %.

Conclusion

In this study, we have established a causal relationship between COVID-19 and the risk of hypopituitarism. This suggests that vigilance against the development of hypopituitarism is warranted following SARS-CoV-2 infection, and underscores the potential for long-term autonomic dysfunction in post-COVID-19 clinical management.

1. Introduction

The emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has contributed to the pandemic known as coronavirus disease 2019 (COVID-19), turning it into the global public health menace of this century. COVID-19 has also been significantly associated with the symptoms of multi-organ system diseases beyond lungs, exhibiting severe complications, including cardiovascular disease (Moneim et al., 2022), coagulation (Yamada and Asakura, 2022), neurological (Gill and Cho, 2023), and endocrine (Rossetti et al., 2022) disorders.

The pituitary gland is the central organ in the neuroendocrine regulatory system. The anterior lobe of pituitary gland, which mainly secretes nine important hormones, is responsible for its functional aspect, while the posterior lobe known as the neurohypophysis, primarily receives and stores two hormones. Hypopituitarism (HP) is characterized by a decline in pituitary or hypothalamic function resulting in inadequate secretion of anterior pituitary hormones (Yeliosof, 2019), which can increase the risk of morbidity and mortality, as well as reduce quality of life. HP is a rare disease with an estimated prevalence of 30–45 cases per 100,000 individuals and an annual incidence of 4–5 cases per 100,000 individual (Schneider et al., 2007). Pituitary damage, resulting from various causes such as hypothalamic and pituitary tumors, inflammation, autoimmunity, vascular injury, genetic abnormalities, irradiation and trauma, is the primary etiology of HP. Previous studies have indicated that Orthohantavirus infection can lead to HP in many individuals (Bhoelan et al., 2019).

Case reports have indicated a temporal correlation between SARS-CoV-2 infection and pituitary inflammation followed by HP, a rare condition characterized by lymphatic, granulomatous, plasmacytoid or yellow verrucous infiltration of the pituitary gland and stalks (Partenope et al., 2023). Retrospective studies have identified a significant correlation between SARS-CoV-2 infection and HP, suggesting COVID-19 as a potential new predisposing factor for pituitary apoplexy (PA) (Laroy and Mcguire, 2021). Several studies have suggested that HP may be associated with COVID-19, indicating the need for further research to research to establish a causal relationship between the two. This hypothesis could be further investigated through Mendelian randomization (MR) studies.

Confirmation causality between COVID-19 and HP is challenging due to potential reverse and confounding causation. While randomized controlled trials (RCTs) are considered the gold standard for investigating exposure-outcome causality, ethical constraints imposed by medical committees, strict population selection criteria, logistical challenges, and limitations in result extrapolation hindered our ability to establish the causal link between COVID-19 and HP through RCTs.MR studies use single nucleotide polymorphisms (SNPs) as instrumental variables (IVs) to infer causality between exposure and outcome variables. Therefore, MR study methods are not prone to confounding or reverse causation, akin to random assignment in RCTs. In this study, we employed a two-sample MR to investigate the causal association between COVID-19 and HP.

MR and colocalization analyses may be employed to determine whether COVID-19 serves as a significant risk factor for increasing the incidence of HP or if a causal relationship exists between the two. This approach lays the groundwork for further investigation into the underlying mechanisms.

2. Methodology

2.1. Study design

MR analyses were conducted using publicly available pooled data from genome-wide association analyses (GWASs) to evaluate the causality between COVID-19 and HP. The exposure and outcome cohorts comprised individuals of European descent to mitigate population stratification bias. All data utilized in this study were obtained from publicly available sources that had obtained consent and ethical approval from the participants involved, thus obviating the need for institutional review board approval. Additionally, this study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology 2021 (STROBE) and STROBE-MR guidelines (Skrivankova et al., 2021). MR analyses and sensitivity analyses were conducted sequentially.

The key assumptions for MR analyses in this study, as depicted in Fig. 1, include: (1) IVs are strongly associated with the exposure; (2) IVs are independent of confounders affecting both exposure and outcome; (3) IVs influence the outcome solely through the exposure.

Fig. 1.

Three Assumptions of MR Study Note: (I) Genetic instrumental variables (IVs) are strongly associated with COVID-19; (II) Genetic IVs do not affect hypopituitarism through confounding factors; (III) Genetic IVs do not directly influence hypopituitarism but only through COVID-19.

2.2. Data sources

We utilized two publicly available pooled datasets from GWAS, one for exposure and the other for outcome. Table 1 presents the profiles of COVID-19 and HP in the GWAS. The COVID-19 associated SNPs datasets (case N = 8316 and control N = 1549,095) were obtained from European Bioinformatics Institute (EBI). The HP associated SNPs datasets (case N = 910 and control N = 211,123) were obtained from FinnGen. Both of the above data populations coming from European origin, are online and publicly available in the Integrative Epidemiologic Unit (IEU) GWAS database (Browse the IEU Open GWAS project (mrcieu.ac.uk).

Table 1.

Profiles of Exposure and Outcomes in GWAS Datasets.

	GWAS ID	Consortium	Year	population	Sex	Sample size		NSNP
						Case	Control
exposure	ebi-a-GCST011082	EBI	2020	European	All	8316	154,9095	6814,406
outcome	finn-b-E4_HYPOPIT	FinnGen	2021	European	All	910	211,123	16,380,448

GWAS ID: genome-wide association study identity; NSNP: number of single nucleotides polymorphism; EBI: European Bioinformatics Institute.

2.3. Screening of instrumental variables

We use the Two-sample MR R package to screen the genetic IVs. The specific conditions of selecting IVs must meet as following: (1) SNPs associated with COVID-19 reached genome-wide significance (P₁<5 ×10^–8); (2) to avoid linkage disequilibrium, R²< 0.001, clumping distance> 10,000 kb; (3) SNPs has nothing to do with the outcome (P₂>5 ×10⁻⁵); (4) to ensure a robust estimated effect of genetic variation, SNPs with F> 10 included. F statistical power was calculated to test genetic effect of IVs, using the formula F = R² (n-2) / (1-R²), where R² is the variance of COVID-19 explained by IVs, and n is the sample size. F statistical power greater than 10 makes IV in better validity. SNPs' main information, including effect allele, other allele, beta, standard error, and P value, was systematically collected for further analysis. When the number of SNPs was less than or equal to 3, adjusted for P1 < 5 × 10⁻⁶, continued R² < 0.001, clumping distance> 10,000 kb, and P₂ > 5 × 10⁻⁵. Outliers were appropriately removed, and SNPs with palindromic alleles (A/T or G/C) were excluded to prevent strand ambiguity problems.

2.4. Mendelian analysis

Four methods, including inverse variance weighting (IVW), MR-Egger regression, weighted median, and weighted mode, were employed in this two-sample MR study (Md, 2009). The results were presented as odds ratios (OR) with corresponding 95 % confidence intervals (95 % CI) and p-values. By leveraging the strengths of each MR method, these four approaches complemented one another. IVW, a standard method for MR, played a pivotal role in establishing causal relationship between COVID-19 and HP. The MR-Egger method was utilized to assess and adjust for pleiotropy in MR analysis. In the weighted median method, a minimum of 50 % of SNPs had to confirm the validity of IVs. The weighted mode method assigned weights to the causal effects of various genetic variations on traits, providing the final estimate while mitigating bias caused by deviations in genetic variation estimates. These four methods are widely acknowledged as scientifically robust approaches to MR. In cases of statistically significant heterogeneity, a random effects model was applied; otherwise, a fixed effects model was employed. A significance level of P < 0.05 indicated a statistically significant difference.

2.5. sensitivity analysis

Sensitivity analysis is a crucial method for assessing potential bias in MR, encompassing heterogeneity and pleiotropy tests. To assess heterogeneity in IVW methods, we computed the Cochran's Q statistic and evaluated heterogeneity significance based on the p-value (where P > 0.05 indicates no heterogeneity). The MR-Egger method, tailored for identifying and correcting for horizontal pleiotropy in MR, employed the MR-Egger intercept to assess whether COVID-19 manifested a polytropic effect on HP (where P > 0.05 indicates no horizontal pleiotropy).

Additionally, the leave-one-out analysis was conducted to identify potentially anomalous SNPs. Funnel plots were used to illustrated the symmetric distribution of selected SNPs. Scatter plots were employed to examine the causality between COVID-19 and HP. Finally, forest plots visually presented the effect size of individual SNPs' of COVID-19 on HP.

2.6. Colocalization analysis

Colocalization analysis was conducted to examine whether COVID-19 and HP shared the same causal variation SNP.

3. Results

3.1. Genetic IVs

SNPs were screened based on the MR package of IVs and corrected after multiple tests, the F statistics of IVs were all greater than 10, indicating the absence of weak instruments. Five SNPs were selected: rs111837807, rs13050728, rs2109069, rs35081325, rs10774679.

3.2. Mendelian randomization analysis

We employed IVW, MR-Egger regression, weighted median approach, and Weighted mode to assess the causal relationship between COVID-19 and HP. The results of MR analysis are presented in Table 2. IVW results (OR=1.283, 95 % CI (1.017, 1.617), p = 0.035) indicated a causal association between COVID-19 and HP. MR-Egger regression (OR=1.355, 95 % CI (0.850, 2.160), p = 0.291). The weighted median approach (OR=1.327, 95 % CI (1.017, 1.732), p = 0.037), while the weighted mode approach (OR=1.330, 95 % CI (0.993, 1.782), p = 0.128). Additionally, there was no reverse causality between HP and COVID-19.

Table 2.

Results of MR Analysis of COVID-19 and HP.

Exposure ID	Outcome ID	method	NSNP	b	se	P	OR（95 % CI）
ebi-a-GCST011082	finn-b-E4_HYPOPIT	IVW	5	0.249	0.118	0.035	1.283(1.017, 1.617)
		MR Egger	5	0.304	0.238	0.291	1.355(0.850, 2.160)
		Weighted median	5	0.283	0.136	0.037	1.327(1.017, 1.732)
		Weighted mode	5	0.285	0.149	0.128	1.330(0.993, 1.782)

IVW: inverse-variance weighted; OR: odds ratio; CI: confidence interval. Statistical significance was set at p < 0.05.

3.3. Sensitivity analyses

To comprehensively assess potential biases in MR studies, we conducted sensitivity analyses, encompassing heterogeneity analysis and horizontal pleiotropy analysis.

3.3.1. Heterogeneity test

Table 3 demonstrates the absence of heterogeneity in the Cochran Q test for MR Egger (Q= 3.348, p = 0.341) and IVW (Q= 3.430, p = 0.489).

Table 3.

Heterogeneity Test Results.

Exposure ID	Outcome ID	method	Heterogeneity--CochranQ
			Q	Q_df	Q_pval
ebi-a-GCST011082	finn-b-E4_HYPOPIT	MR Egger	3.348	3	0.341
		IVW	3.430	4	0.489

IVW: Inverse variance weighted. P > 0.05 for no heterogeneity.

3.3.2. Horizontal pleiotropy validity test

MR-Egger and MR-PRESSO were incorporated in the horizontal pleiotropy analysis. As depicted in Table 4, no significant intercept (intercept=- 0.015, p = 0.804) and no horizontal multivariate validity was observed. MR-PRESSO showed no outliers, p = 0.085.

Table 4.

Results of Horizontal Pleiotropy Test.

outcome	exposure	MR-Egger		MR-PRESSO
		intercept	p	Outliers	p
Hypopituitarism || id: finn-b-E4_HYPOPIT	COVID−19 (hospitalized vs population) RELEASE 5 || id: ebi-a-GCST011082	−0.015	0.804	NA	0.085

p > 0.05 in MR-Egger representing no horizontal multiplicity. NA: mean there was no outliers.

3.4. Funnel plot, scatter plot, forest plot

In the Fig. 2, A, B, and C presented the scatter plot, funnel plot, forest plot respectively, depicting the relationship between COVID-19 and HP. As shown in Fig. 2-A directly depicted the causality of COVID-19 on HP through the scatter plot. Additionally, Fig. 2-B, the funnel plot demonstrated a symmetric distribution of all selected SNPs. Fig. 2-C displayed the forest plot, showcasing the effect size of each SNP of COVID-19 on HP.

Fig. 2.

The relationship between COVID-19 and HP Note: A, B, and C presented the scatter plot, funnel plot, forest plot respectively, D showed colocalization result between COVID-19 and HP.

3.5. Colocalization results

Fig. 2-D illustrated the colocalization relationship between COVID-19 and HP. The majority of colocalization associations had a posterior probability (PP.H4.abf) exceeding 80 %. However, the colocalization probability between COVID-19 and HP in this study was determined to be PP.H4.abf= 9.55 %. These results suggest no shared genetic factor between COVID-19 and HP. Combining findings from MR and colocalization studies, COVID-19 emerges as a significant causal factor in elevating the risk of HP.

4. Discussion

This study represents a pioneering effort in utilizing MR and Colocalization methods to elucidate causality and underlying mechanisms in the relationship between COVID-19 and HP. The study successfully identified and validated a causal relationship using IVW and Weighted median, integrating GWAS data on COVID-19 and HP. Simultaneously, sensitivity analyses verified the stability of the results. The colocalization analysis further clarified the correlation between COVID-19 and the risk of HP.

Alongside some known risk factors as hypertension, anticoagulation, pregnancy, surgery, Ragni et al. (2024) found COVID-19 infection might represent an predisposing factor for PA. presently, numerous studies have identified endocrine dysfunction following viral infections, with particular attention being given to the HP in COVID-19 infection. Several studies have suggested a potential association between the two. In one case report reversible hypopituitarism due to hypophysitis developed after COVID-19 infection (Gorbova et al., 2022). Hypophysitis was identified following COVID-19 infection, with anterior or posterior pituitary dysfunction observed in 85 % of patients (Menotti et al., 2024). Neuroendocrine changes are observed following COVID-19 (Bellastella et al., 2022), with patient frequently displaying hypothalamic-pituitary dysfunction (Alzahrani and NohaAljomaiah, 2021). COVID-19 may serve as a risk factor for the development of PA (Martinez-Perez et al., 2021). Pre-existing pituitary diseases may exacerbate COVID-19 clinical symptom and severity, with HP dysfunction frequently coexisting with complications such as hypertension, hyperglycemia, and obesity. These complications are associated with heightened adverse outcomes and mortality rates among patients following SARS-CoV-2 infection. Direct or indirect pituitary damage poses a potential risk associated with COVID-19 (Frara et al., 2021). These findings demonstrate an association between COVID-19 and HP, aligning with our MR findings indicating a causality between COVID-19 and HP, and potentially neuroendocrine disease.

The pathogenesis of this disease remains unclear and necessitates further investigation: both SARS and COVID-19 can impact the hypothalamic-pituitary-adrenal (HPA) axis, resulting in reversible PA or direct hypothalamic damage, ultimately leading to hypothalamic-pituitary dysfunction (Pal, 2020). A study found that COVID-19 may impact pituitary function, particularly the HPA and GH axes, primarily due to hypoxia, endothelial dysfunction and autoimmune changes induced by the SARS-CoV-2 infection (Emre Urhan et al., 2022). It involves inflammation, cytokine overproduction, and cell death, processes associated with the overproduction of reactive oxygen species and the induction of oxidative stress (Giunta et al., 2024). The virus may enters the central nervous system (CNS) and interacts significantly with angiotensin-converting enzyme receptors by binding to spike proteins in the capillary endothelium and directly damaging pituitary cells (Menotti et al., 2023).

Notably, pituitary dysfunction and central adrenal insufficiency have been documented in COVID-19 patients even without hypophysitis. Emerging evidence suggests autoimmunity may drive this phenomenon, potentially representing a distinct phenotype. In a study, four COVID-19 patients result in adrenocortical insufficiency, demonstrating the presence of antipituitary in three and antihypothalamic antibodies in one, pituitary and hypothalamic autoimmunity may significantly contribute to the neuroendocrine abnormalities associated with COVID-19 (Gonen et al., 2022). Host immune evasion is also a significant pathogenic mechanism. Pérez et al. (2022) demonstrated a 60 % prevalence of anti-ACTH IgE-class antibodies in critically ill COVID-19 patients with adrenal insufficiency. These findings suggest SARS-CoV-2 utilizes molecular mimicry to disrupt neuroendocrine-immune crosstalk, evading host defenses via dual mechanisms: primary immune blockade and HPA axis suppression.

Studies have also found effects of COVID-19 on the hypothalamic-pituitary-gonadal axis. Individuals who have contracted COVID-19 are more likely to compromise male fertility due to the susceptibility of the male reproductive system (Kalfas et al., 2023). A cohort study discovered that COVID-19 could impact the level of sex hormones, including testosterone, thereby potentially affecting male reproduction (Lan et al., 2024). A study involving 1250 patients with acute or recent COVID-19 infection found significantly lower sperm quality compared to controls. Additionally, patients exhibited lower levels of total testosterone and follicle-stimulating hormone (Cannarella et al., 2024).

Patients who have experienced COVID-19 may develop post-acute phase complication, result in heterogeneous negative impacts across multiple systems, a condition known as the "long COVID." (Yang et al., 2022). COVID-19 or Long COVID, after infection with SARS-CoV-2, is comorbid with neuroendocrine-related diseases in many patients. The Long COVID syndrome is characterized by a series of debilitating conditions that are associated with symptoms of autonomic dysfunction, lower cortisol levels, reduced HPA axis activity, and increased pro-inflammatory conditions (Giunta et al., 2024). The mechanisms of neuropsychiatric complications in long COVID are multifactorial, involving tissue damage from direct CNS viral involvement, systemic inflammation and oxidative stress. Long COVID syndrome reduces the quality of life and increases the risk of developing the most common age-related diseases. These findings have attracted the attention of scholars. A multitude of case reports and cohort studies have demonstrated a consistent correlation with that of this study, confirming that COVID-19 increased the risk of HP. Building upon the established causal relationship between COVID-19 and HP in this study, further investigation into the mechanisms governing the onset of HP in individuals with COVID-19 is warranted.

In this investigation, data were obtained from FinnGen and EBI, ensuring no data overlap. Rigid selection criterion, including p < 5 × 10^-⁶ or p < 5 × 10^-⁸, R² < 0.001, clumping > 10,000 kb, and F-statistical validity greater than 10, were applied to IVs to mitigate confounding and enhance their robustness. Detection and correction of bias due to pleiotropy were performed using MR-Egger regression. This pioneering study investigated the causal relationship between COVID-19 and HP using MR, revealing an increasing risk of HP associated with COVID-19.

However, there are several limitations that should be acknowledged. Firstly, the study was confined to a population of European origin, making it challenging to generalize the findings to other populations. Secondly, we were unable to determine whether a nonlinear or U-shaped relationship existed between COVID-19 and HP. Thirdly, the study acknowledged the challenge posed by developmental compensation, which may affect the accuracy of the MR results, a circumstance that is inherently unavoidable.

In conclusion, the current MR analyses suggest a causal relationship between COVID-19 and increased risk of HP. This suggests that vigilance for the development of HP is necessary following SARS-CoV-2 infection, as HP or even long COVID syndrome autonomic dysfunction may become focal points of clinical chronic disease management in the future. Further studies are required to explore the precise mechanisms and confirm the potential clinical value of this preventive and therapeutic strategy.

5. Conclusion

By utilizing SNPs associated with COVID-19 from GWAS data as IVs, this MR study offer robust evidence supporting a causality between COVID-19 and the risk of HP in the European population. While these findings are compelling, additional RCTs are necessary to validate these results. Furthermore, this study paves the ways for future research to investigate the mechanisms underlying HP following COVID-19 infection, potentially leading to the development of novel treatments aimed at mitigating HP development in individuals with COVID-19 and enhancing overall quality of life.

CRediT authorship contribution statement

Meichen Tian: Writing – original draft, Software, Methodology, Formal analysis, Data curation, Conceptualization. Xiyang Li: Writing – original draft, Supervision, Resources, Methodology, Formal analysis, Data curation, Conceptualization. Guifang Yan: Validation, Software, Resources, Methodology. Xi Wang: Validation, Software, Resources, Methodology, Investigation, Formal analysis. Qinxuan Li: Writing – review & editing, Visualization, Validation, Supervision, Project administration, Methodology, Investigation, Conceptualization. Li You: Validation, Software, Methodology, Formal analysis, Data curation. xiao jiujia: Writing – review & editing, Visualization, Supervision, Resources, Project administration, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Ethics statement

All data used in this study were publicly available from studies that received consent and ethical approval from the participants involved. In accordance with legislation and institutional policies, ethical approval and review were deemed unnecessary for this study involving human participants.

Funding

None

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The author is an Editorial Board Member/Editor-in-Chief/Associate Editor/Guest Editor for this journal and was not involved in the editorial review or the decision to publish this article.