Whole-Blood Expression of Candidate Genes Linked with Pheochromocytoma in Post-Surgery Patients: A Pilot Study

Abstract

Background: Pheochromocytomas (PCCs) and paragangliomas (PGLs), collectively known as PPGLs, are rare neuroendocrine tumors that produce catecholamines. The majority of PPGL cases are caused by germline and/or somatic mutations in over 20 different genes. A study of post-surgical PCC patients revealed a high risk of new tumor recurrence in both hereditary and apparently sporadic cases, suggesting that some germline mutations remain undetected. Since transcript levels can indicate gene dysfunction, our study focuses on the transcriptional profiling of PCC-associated genes in post-surgical patients. Methods: RT-PCR was performed on blood samples from patients and a control group. The t-SNE algorithm was applied to the transcriptional data. Sanger sequencing was used to identify mutations in the coding sequences of the VHL, SDHB, RET, and NF1 genes. Results: We obtained transcriptional profiles for 11 genes involved in the Krebs cycle and for 21 genes involved in the hypoxia, PI3K/AKT/mTOR, and RAS/RAF/ERK signaling pathways. We identified a minimal set of 16 genes with stable transcription levels that can be used to differentiate PCC patients from controls. Germline mutations in the VHL, SDHB, RET and NF1 genes, which correlated with an altered transcriptional profile, were detected in three patients. Conclusions: Our pilot data suggest that transcript levels of the genes involved in Krebs cycle, hypoxia, PI3K/AKT/mTOR, and RAS/RAF/ERK signaling pathways indicate their potential suitability as a candidate diagnostic marker. The results from this pilot study form the basis for a larger project to investigate gene transcription in an expanded cohort of patients who have undergone surgery for PCC.

1. Introduction

Pheochromocytomas (PCCs) and paragangliomas (PGLs), collectively known as PPGLs, are rare neuroendocrine tumors producing excessive amounts of catecholamines. Pheochromocytomas (PCCs) arise from chromaffin cells of the adrenal medulla and account for approximately 80–85% of all catecholamine-secreting neuroendocrine tumors, with the remaining 10–15% being extra-adrenal paragangliomas, commonly found in the head and neck region [1,2,3,4]. PCCs are also known to be associated with genetic syndromes such as multiple endocrine neoplasia type 2 (MEN-2), von Hippel–Lindau disease (VHL), or neurofibromatosis type 1 [2]. According to the current classification proposed by the World Health Organization, all PPGLs are considered potentially metastatic [3,4].

Most PPGLs are caused by germline and/or somatic mutations, with germline mutations being present in 30–40% of PPGLs, numbers that are several times higher than those in the case of other malignant diseases [5]. It was proposed that PPGLs should be divided into three types, depending on the gene cluster whose dysfunction leads to the disease: metabolic, tyrosine kinase, and Wnt-dependent [6]. In sum, all three clusters included more than 30 genes. Thus, PPGLs are genetically heterogeneous diseases caused by mutations in various interconnected genes that form a single regulatory network. Due to this interconnectedness, dysfunction in even a single gene within the network can disrupt the entire system and lead to tumor development.

Since PPGL is a malignant disease, its occurrence can be explained by Knudson’s two-hit tumor hypothesis, as well as other malignancies. According to this hypothesis, the transformation of a normal cell into a tumor requires two sequential mutational events—“hits”. In the case of hereditary cancer, the first hit is a germline mutation, leading to the formation of a cell with a high risk of malignant transformation. The transformation occurs as a result of the second hit—a somatic mutation. Sporadic cancer is less common and is the result of two mutations in a somatic cell. According to current concepts, three to six additional genetic mutations (depending on the nature of the initial or predisposing mutation, which can predetermine the disease pathway) are required to complete the tumor formation process [7].

According to a large-scale study of 173 PPGL tumor samples, approximately one-third of patients harbored germline mutations affecting gene coding sequences, indicating a hereditary origin for these cases [8]. However, the incidence of hereditary forms is likely underestimated, as genetic screening typically focuses on the exome. Functionally disruptive mutations can occur not only in coding regions but also in regulatory elements, which are often poorly characterized. Furthermore, the pathogenicity of identified variants cannot always be unequivocally determined. This limitation is significant given that over 90% of disease-associated genetic variants are located in noncoding regions of the genome [9].

The fact that a substantial number of germline mutations remain undetected is indirectly supported by clinical follow-up data. A study of postoperative PPGL patients showed that the long-term risk of new tumor recurrence was significantly higher in those with confirmed hereditary tumors. However, a considerable risk persisted even in patients with apparently sporadic variants (20-year risk: 38% vs. 6.5%, respectively) [10].

Gene transcription analysis serves as a useful tool for detecting mutations. Changes in transcript levels can indicate loss-of-function or gain-of-function mutations arising from alterations in either the gene’s coding sequence or its regulatory regions, such as transcription factor binding sites or enhancers [11,12,13]. The transcription level can serve as an indicator of epigenetic changes affecting transcription. In addition, intermediate metabolites of the Krebs cycle—such as oxaloacetate and α-ketoglutarate, which can accumulate as a result of mutations—can act as inducers of transcription for genes involved in cellular signaling pathways [14]. Four gene expression subtypes were identified in PPGLs: kinase signaling, pseudohypoxia, Wnt-altered, and cortical admixture [8].

Blood is the most convenient tissue for noninvasive transcriptional testing. Numerous studies have demonstrated a correlation between the expression of marker genes in the blood cells of patients and various cancer types (colorectal cancer, renal cell carcinoma, lung cancer, breast cancer, and stomach cancer). This correlation forms the basis for developing targeted gene expression panels to predict disease severity and treatment response [15,16,17,18,19,20]. Recently, blood transcriptome profiling has been used for the early identification of Suboptimal Health Status (SHS) [21]. The pattern of differentially expressed genes was used as a predictive transcriptomic biomarker to identify SHS in individuals who subjectively felt healthy but had alterations at the subcellular level.

Analysis of the blood transcription profiles from PCC patients in remission may help identify the initial driver mutation (“first hit”). In our study, we examined the blood transcript levels of 32 candidate genes known to be associated with PPGLs in patients who had undergone tumor surgery.

2. Materials and Methods

Study design. This is a single-center non-interventional observational case–control study.

Samples. The patient sample included 12 patients (6 women and 6 men) aged 22–71 years (median 49.5 [40.5–59.5]). We recruited patients with verified PCC based on immunohistochemical investigation after surgical removal of the adrenal tumor within the previous year. Thus, due to very low PCC incidence, only twelve patients were included in our study. The matched control sample (n = 20) consisted of healthy individuals without arterial hypertension, no history of oncological/endocrine diseases, and no other clinical signs of diseases that could be associated with PCC: 10 women, 10 men, aged 24–63 years (median 45.0 [39.5–61.0]). The samples did not significantly differ by gender (p = 0.978) or age (p = 0.893). Venous blood samples were collected on an empty stomach and transferred to the laboratory for RNA isolation within 1 h of collection. All blood samples were obtained at the M.F. Vladimirskiy Moscow Regional Research Clinical Institute.

Clinical methods. Daily urinary excretion of fractionated metanephrines and normetanephrines was measured using high-performance liquid chromatography–tandem mass spectrometry (HPLC-MS/MS). The reference values were <320 μg/day for metanephrines and <390 μg/day for normetanephrines. CT was performed using Aquilion Prime Canon (Toshiba), 160-slice, with a contrasting agent (Omnipaque 350), according to the standard protocol. Data on the size and density were collected. The volume of adrenal tumors was calculated using the following formula: volume (cm³) = (length, mm) × (width, mm) × (thickness, mm) × 0.0005. Immunohistochemical succinate dehydrogenase staining was performed on 4 μm thick formalin-fixed, paraffin-embedded (FFPE) adrenal tissue sections using Epredia Autostainer 720 (Shandon Diagnostics Limited, Runcorn, UK), with a Dako EnVision FLEX High pH kit (Dako Omnis, Agilent Technologies, Santa Clara, CA, USA). A rabbit monoclonal antibody against SDHB (clone YN02535r; Wuhan Elabscience Biotechnology Co., Ltd., Wuhan, China) was used at a dilution of 1:200.

RNA isolation and transcription analysis. RNA was isolated from whole blood using the Extract RNA reagent (Evrogen, Moscow, Russia) according to the manufacturer’s protocol. RNA concentration was measured using a Nanodrop ND-2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). DNA was purified from RNA using DNase I (Thermo Fisher Scientific) according to the manufacturer’s protocol. cDNA was synthesized using the isolated RNA template using the MMLV RT kit (Evrogen), according to the manufacturer’s instructions. Random primers were used to initiate cDNA synthesis on the RNA template. Real-Time PCR was performed using a MiniOpticon Real-Time PCR System amplifier (Bio-Rad Laboratories Inc., Hercules, CA, USA). PCR was performed using qPCRmix-HS mixture (Evrogen). PCR had the following scheme: denaturation at 95 °C for 10 s, primer annealing at 55 °C for 30 s, and elongation at 72 °C for 1 min. The study included RT-PCR transcription analysis of the following genes: ACO1, IDH1, DLST, SDHA, SDHB, SDHC, SDHD, FH, MDH2, GOT2, SLC25A11, EGLN1, EGLN2, EPAS1, DNMT3A, VHL, RET, MET, MERTK, FGFR1, NGFR, KRAS, HRAS, RAF1, BRAF, AKT1, PI3CA, MTOR, MAPK1, MAX, TMEM127, and RPS6KB1. The ACTB, M2B, RPL13A, and YWHAZ genes were used as references. Primer sequences are presented in Table S1.

To select a reference gene, we analyzed the transcription of four standard reference genes (ACTB, M2B, RPL13A, and YWHAZ) in the control sample of 20 individuals. To assess the contribution of technical error, the transcription of each gene in each individual was analyzed in triplicate. The SD of Ct within the three technical replicates averaged 0.17, 0.15, 0.11, and 0.15 for ACTB, M2B, RPL13A, and YWHAZ, respectively. The SD of Ct for the entire sample was 0.9, 0.7, 1.1 and 1.4 for ACTB, RPL13A, M2B, and YWHAZ, respectively. For normalization, geometric mean of Ct for ACTB and RPL13A genes was used as they showed the best stability in the GeNorm testing (https://genorm.cmgg.be/, accessed on 5 November 2025).

To assess target gene stability, the mean, median, standard deviation (SD), and coefficient of variability (SD/mean) were calculated. The following scale was used to evaluate the variability criterion: CV < 1—high stability; 1< CV < 2—medium stability; CV > 2—low stability.

Statistical analysis of transcription data. The transcription levels of the selected genes were analyzed separately for the control and patient groups using the delta-Ct method. Expression levels of the studied genes were calculated by normalizing to the reference gene. Expression of each gene was analyzed in 3 replicates. Statistical analysis was performed using StatTech v. 4.2.6 (StatTech LLC, Kazan, Russia). The normality of the data distribution was tested using the Shapiro–Wilk test. Fold change in the median transcription level was calculated. Comparison of the two groups (patients and controls) for quantitative indicators of gene expression, whose distribution deviated from normal, was performed using the Mann–Whitney U-test with Bonferroni and Benjamini–Hochberg (FDR) multiple alignment corrections. Differences were considered statistically significant at p-value < 0.05.

Correlation analysis. To analyze correlations between gene transcription levels, we used the Spearman correlation, which is a nonparametric measure of the monotonic relationship between two variables based on data ranks, does not require a normal distribution, and is robust to outliers. For correlation analysis, the online resource SRplot was used (https://www.bioinformatics.com.cn/, accessed on 1 October 2025).

Distances calculation and data visualization. A joint analysis of patient and control data was conducted using a pairwise Euclidean distance matrix. The coordinates were represented by two sets of genes for which the Mann–Whitney U-test confirmed statistically significant differences between patients and controls after two different multiple comparison corrections: Bonferroni and FDR. The data were standardized to ensure that the contribution of each gene was equal. The nonlinear dimensionality reduction algorithm t-SNE [22] was applied to the patients’ and controls’ transcriptional profiles to visualize the data on a plane. This method preserves the local structure of the data, meaning that close points remain close during visualization, facilitating clustering analysis. Since t-SNE can be unstable with small datasets, the main conclusions were drawn using the distance matrix.

Genotyping. To analyze the NF1, VHL, RET, and SDHB gene sequences, PCR was performed on DNA templates. PCR fragments of the genes were sequenced using the Sanger method (Evrogen). Chromatograms were analyzed using UGENE software (version 51.0, UNIPRO LLC, Novosibirsk, Russia). Information on the clinical significance of the identified variants was found in the ClinVar database (https://www.ncbi.nlm.nih.gov/clinvar/, accessed on 1 December 2025) and dbSNP (https://www.ncbi.nlm.nih.gov/snp/, accessed on 1 December 2025) on the NCBI server. For previously undescribed mutations, the impact of the mutation on protein function and structure was analyzed using the PolyPhen-2 tool [23]. Mutation pathogenicity analysis was performed using the MutationTaster 2025 tool [24].

3. Results

3.1. Analysis of Gene Transcription Levels of Genes Associated with PPGL in Postoperative Patients with PCC

Blood for transcriptional analysis was drawn at a single timepoint, approximately one year after surgery (within a window of 10–14 months). This timepoint was chosen to avoid the acute inflammatory and metabolic postoperative period while assessing a presumed “baseline” state of predisposition. The study’s aim was not to correlate transcriptional profiles with long-term recurrence but to test for the presence of a stable transcriptional signature in remission. The main clinical characteristics of the patients at diagnosis are presented in

Table 1. Main clinical characteristics of the analyzed patients with a PCC.

Patient Number	Age *	Gender	Uni- (1) or bi-(2) Lateral Tumor	Adrenal Tumor		Fold Change Relative to Reference Value		SDHB Expression **	Concomitant Diseases
				Max Size, mm	Volume, cm3	MN	NMN
1	42	m	1	26	13.2	NC	2.5	no	no
2	55	m	2	38	38.6	6.0	3.0	no	Medullar thyroid cancer
				20	4.1
3	36	f	1	47	91.0	NC	43.6	yes	no
4	44	f	1	18.5	4.7	2.7	NC	no	no
5	61	f	1	50	15.2	27.4	17.4	yes	Primary hyperparathyroidism, Non-functional adenoma of another adrenal gland (13 × 8 mm), Renal cell carcinoma
6	43	m	1	70	276.4	35.7	11.9	yes	no
7	39	m	1	46	60.4	3.2	2.7	yes	no
8	55	m	1	39	23.8	NC	6.46	yes	no
9	63	m	1	50	115.2	10.2	5.7	yes	no
10	22	f	2	50	86.1	NC	13.3	yes	Pancreatic non-functional tumor (18 × 12 × 14 mm)
				16	3.3
11	58	f	1	20	4.1	NC	4.9	no	no
12	71	f	1	83	389.4	2.3	7.7	yes	Neurofibromatosis type 1, ovarian androgen-producing tumor

Legend: *—age at diagnosis, MNs—metanephrines, NMNs—normetanephrines, NC—not changed, **—SDHB expression at the adrenal tumor tissue.

. Eleven of these patients (except for patient 12, who experienced persistent disease) showed no recurrence one year after tumor removal.

For our analysis, we selected 32 genes from Clusters 1 and 2, as classified by Nolting et al. [6]. Cluster 1 includes genes involved in the tricarboxylic acid cycle (ACO1, IDH1, DLST, SDHA, SDHB, SDHC, SDHD, FH, MDH2, GOT2, SLC25A11) and the HIF-α hypoxia signaling pathway (EGLN1, EGLN2, EPAS1, DNMT3A, VHL), where loss-of-function mutations are most common. Such mutations can lead to decreased expression of co-regulated genes within the cluster. Cluster 2 includes genes involved in the PI3K/AKT/mTOR and RAS/RAF/ERK tyrosine kinase signaling pathways (EGLN1, EGLN2, EPAS1, DNMT3A, VHL) (RET, MET, MERTK, FGFR1, NGFR, KRAS, HRAS, RAF1, BRAF, AKT1, PI3CA, MTOR, MAPK1, MAX, TMEM127, RPS6KB1), which are frequently affected by gain-of-function somatic mutations. These mutations result in the transcriptional activation of downstream targets, stimulating cell growth and linking Cluster 2 dysfunction not only to PPGLs but also to other cancer types. All genes are expressed in both adrenal cells and whole blood (https://gtexportal.org/home/, accessed on 1 November 2025).

As transcription was analyzed using RT-PCR, the critical methodological steps were the selection of stable reference genes and the assessment of gene expression variability in control samples. We evaluated the expression stability of four candidate reference genes (ACTB, RPL13A, M2B, and YWHAZ) using the GeNorm algorithm. Based on this analysis, ACTB and RPL13A were selected for normalizing target gene expression.

For each gene in the control set, we calculated the standard deviation (SD), interquartile range (IQR), and coefficient of variation (CV) for the normalized expression values. The stability measures for all genes are presented in

Table 2. Gene transcription stability in control samples.

Gene	ACO1	IDH1	DLST	SDHA	SDHB	SDHC	SDHD	FH	MDH2	GOT2	SLC25A11	EGLN1	EGLN2	EPAS1	DNMT3A	VHL
CV	0.48	0.58	0.76	0.54	0.85	2.40	0.89	0.63	0.75	1.11	1.75	1.38	0.96	0.74	2.84	0.68
Stability	High	High	High	High	High	Low	High	High	High	Mod.	Low	Mod.	High	High	Low	High
Gene	RET	MET	MERTK	FGFR1	NGFR	KRAS	HRAS	RAF1	BRAF	AKT1	PI3CA	MTOR	MAPK1	MAX	TMEM127	RPS6KB1
CV	3.33	0.81	1.31	1.67	3.58	1.15	1.74	7.14	1.06	0.75	1.26	0.57	0.78	1.16	1.90	1.35
Stability	Low	High	Mod.	Mod.	Low	Mod.	Mod.	Low	Mod.	High	Mod.	High	High	Mod.	Low	Mod.

Legend: CV—coefficient of variation; Mod.—moderate.

.

The expression of the genes in Cluster 1 was more stable than that of the genes in Cluster 2. Specifically, 11 of the 16 Cluster 1 genes could be classified as highly stable, compared to only 4 of the 16 genes in Cluster 2. This inherent variability in expression stability must be considered when comparing gene expression between control and patient samples.

Then, we investigated the transcription of the studied genes in the patient sample, analyzed the differences in transcription levels (FoldChange), and assessed the significance of differences between patients and the control group using the Bonferroni and Benjamini–Hochberg (FDR) multiple alignment corrections (

Table 3. Significance of differences in gene transcription between patient and control samples.

Genes	ACO1	IDH1	DLST	SDHA	SDHB	SDHC	SDHD	FH	MDH2	GOT2	SLC25A11	EGLN1	EGLN2	EPAS1	DNMT3A	VHL
p_value	0.00005	0.07655	0.00004	4.79 × 10−6	0.00037	0.00002	0.00002	0.26727	0.00043	0.00015	0.15539	0.00256	0.00017	0.00682	0.00173	3.29 × 10−6
FoldChange	2.12	1.59	7.88	28.39	4.31	23.78	5.39	1.24	4.41	4.28	14.26	6.56	34.31	3.84	9.41	33.61
p_value_adj_B	0.0015	1	0.0013	0.0002	0.0118	0.0005	0.0008	1	0.0137	0.0047	1	0.0818	0.0055	0.2184	0.0553	0.0001
FDR	0.0002	0.1065	0.0002	0.0001	0.0009	0.0001	0.0002	0.3055	0.0009	0.0005	0.1989	0.0043	0.0005	0.0104	0.0031	0.0001
Genes	RET	MET	MERTK	FGFR1	NGFR	KRAS	HRAS	RAF1	BRAF	AKT1	PI3CA	MTOR	MAPK1	MAX	TMEM127	RPS6KB1
p_value	0.00538	0.17930	0.11492	0.02786	0.00043	0.19224	0.00100	0.00173	0.65442	0.49575	0.00011	0.00004	3.29 × 10−6	0.00043	0.36033	0.34026
FoldChange	0.10	2.17	0.56	5.96	9.23	0.41	6.01	11.96	0.65	0.94	17.79	7.75	23.91	4.74	1.83	2.50
p_value_adj_B	0.1723	1	1	0.8915	0.0137	1	0.0322	0.0553	1	1	0.0034	0.0013	0.0001	0.0137	1	1
FDR	0.0086	0.2207	0.1532	0.0405	0.0009	0.2278	0.0020	0.0031	0.6544	0.5117	0.0004	0.0002	0.0001	0.0009	0.3844	0.3755

FoldChange—fold change of median transcription level; p_value_adj_B—p-value with Bonferroni correction; FDR—false discovery rate, Benjamini–Hochberg correction.

and Table S2).

The patient sample as a whole has an increased transcription level of most genes. The Benjamini–Hochberg (FDR) correction resulted in 22 genes, 13 of which were related to Cluster 1, and 9 were genes of Cluster 2. A more stringent Bonferroni correction retained 16 genes, 10 from Cluster 1 and 6 from Cluster 2.

We next analyzed the transcription profiles of Cluster 1 and Cluster 2 genes in individual patients, comparing them to the median expression levels in the control group (Figure 1). The analysis revealed that the expression levels for most Cluster 1 genes were elevated in many patients. In contrast, the increased expression of Cluster 2 genes was observed only in a subset of patients, notably in patients 1 and 2. As shown in Figure 1, elevated transcription of genes involved in the PI3K/AKT/mTOR signaling pathway was also detected in some patients.

Markedly, among the four Cluster 2 genes identified as having highly stable expression, only MET showed a significant difference from the controls only in patients 5 and 12. Significant differential expression of AKT1 was found only in patients 1, 2, 3, and 5. In contrast, MTOR and MAPK1 expression differed significantly from controls in nearly all patients.

Next, we analyzed the correlations between gene transcription levels (Figure 2). We hypothesized that the presence of mutations in patients would impact both the transcriptional profile of individual genes and the overall transcriptional level of the analyzed genes in Clusters 1 and 2, which are part of a single gene network.

In the patient sample, the correlations were stronger than in the control sample. Despite the small size of the studied samples, significant positive correlations between gene transcription both within clusters and between clusters were observed in the patient sample. Strong intercluster transcriptional correlations persisted for the VHL, EGLN1/2, SDHA, and SDHB genes.

Two sets of genes were identified, showing statistically significant (p < 0.05) differences between patients and controls—one with FDR correction and the other with Bonferroni correction. The first set (after FDR correction) included 22 genes: ACO1, DLST, SDHA, SDHB, SDHC, SDHD, MDH2, GOT2, EGLN1, EGLN2, EPAS1, DNMT3A, VHL, RET, FGFR1, NGFR, HRAS, RAF1, PI3CA, MTOR, MAPK1, MAX. The second set (after Bonferroni correction) included 16 genes: ACO1, DLST, SDHA, SDHB, SDHC, SDHD, MDH2, GOT2, EGLN2, VHL, NGFR, HRAS, PI3CA, MTOR, MAPK1, MAX. Because PCC is a genetically heterogeneous disease, transcription profiles vary among individual patients. We therefore compared individual patient transcription profiles based on identified gene sets using the distance matrix (Figure 3) and obtained similar results. As expected, genes selected with the FDR correction allowed us to identify more slight differences between individuals. Patients 3, 4, 6, 7, and 9 were closest to the control sample. Patients 1, 2, 5, and 12 differed most significantly from other patients and controls.

To visualize the results, we employed t-SNE, a nonlinear dimensionality reduction algorithm well-suited for this purpose. The algorithm was applied to two datasets, each obtained after using a different multiple comparison correction method: FDR and Bonferroni. We used various values for the perplexity parameter based on the recommendations in the aforementioned original paper [22]: values between 5 and 50. Since t-SNE is stochastic, different values of a random state (e.g., 0, 20, 42) were used to confirm the stability of the clustering. As an example, Figure 4 shows a t-SNE visualization with perplexity 6 and random state 0. All results are given in Figure S1. Such a choice of the perplexity value is consistent with the distance matrix.

The clustering obtained using the more stringent Bonferroni-corrected gene set appeared more distinct and less noisy, potentially highlighting core differences between patients and controls. The visualizations revealed two distinct groups: the first group includes controls; the second group includes only patients. The resulting visualization largely matches the data description using the pairwise distance matrix.

3.2. Genotyping of Patients

While RNA sequencing is the gold standard for mutation detection, it cannot reliably identify many functionally relevant mutations in noncoding regions. Therefore, we supplemented our analysis with data on the most common PCC-associated mutations, as documented in public databases and the literature. These include primarily germline mutations in Krebs cycle genes. In contrast, mutations in kinase pathway genes are usually somatic, with single germline mutations most frequently found in RET and NF1.

Because germline mutations in the SDHB, SDHD, VHL, RET, and NF1 genes are most common in PPCs [8], we sequenced the coding sequences of these genes in all patients. As a result, patient 2 was found to have two uncharacterized earlier pathogenic mutations in the SDHB gene: c.565T>A (p.Cys189Ser) and c.692T>A (p.Leu231Gln). PolyPhen-2 predicts pathogenicity for both mutations with 100% certainty. Also, patient 2 was found to have in-frame indel mutation in the RET gene, NM_001406743.1:c.1895_1918delinsTGCGGC. The mutation leads to the deletion of two amino acids; one of them is Cys634. Deletions or mutations at Cys634 (like C634R, C634Y) are associated with a more aggressive MEN2 phenotype, including larger tumors, lymph node spread, and worse survival [25]. Patient 4 was found to have a mutation in the VHL gene, c.238A>G (p.Ser80Gly). This mutation has already been described as pathogenic and has been observed previously in patients with PCC [26,27]. Patient 5 was found to have a VHL mutation, c.37_304del, a deletion of 304 nucleotides (alternative transcript ENST00000713815.1). The resulting transcript lacks most of exon 1, which encodes the HIF1A protein recognition site, resulting in loss of function. Patient 12 was found to have a frameshift mutation in the NF1 gene—c.998-999 ins A (p.Y333Term). No mutations in the SDHB, VHL, RET, or NF1 genes were detected in the remaining patients.

4. Discussion

Pheochromocytomas and paragangliomas have one of the highest rates of hereditary predisposition among all tumors, a finding explained by their genetically heterogeneous nature. A number of genes implicated in PPGL pathogenesis belong to a single interconnected network, encompassing both metabolic pathway (e.g., the tricarboxylic acid cycle) and signaling cascades. One approach to identifying dysfunction within this network is the analysis of transcription profiles.

To exclude the potential confounding effect of the tumor on gene expression patterns, our study focused on patients who had undergone tumor resection. This pilot investigation aimed to identify the initial (“first hit”) molecular event in carcinogenesis through transcriptional analysis. Alterations in the expression profiles of PPGL-associated genes in blood cells may reflect an underlying predisposition to the disease.

A large-scale study [8] analyzed 173 PPGL samples with clinically diverse forms of PPGLs and found pathogenic germline and somatic point mutations as well as genomic rearrangements in 89% of samples. Eleven percent of samples showed no identifiable driver mutations, highlighting the need for more comprehensive profiling, including transcriptomic data.

In our cohort, genotyping revealed germline mutations in the coding regions of known susceptibility genes in only 4 out of 12 patients, which aligns with the expected prevalence. However, all patients without exception exhibited certain deviations from the transcriptional patterns of the control group and loss of transcriptional correlations in several genes, indicating a systemic imbalance in PPGL gene network regulation.

The transcriptional impact of specific mutations was evident. For instance, we identified two distinct VHL mutations in the gene—a missense variant and a deletion—in patients 4 (with isolated PCC) and 5 (with some other components of VHL syndrome, see Table 1). These mutation types’ association with different clinical phenotypes has been described before: specific missense mutations often cause isolated pheochromocytoma by selectively stabilizing HIF-2α, while complete loss-of-function deletions lead to VHL syndrome [28,29]. Accordingly, the transcriptional profiles of the patients with different VHL mutations reflected these distinct pathogenic processes. Interestingly, the VHL and SDHx genes are part of the same “pseudohypoxia” cluster, as mutations in both converge on HIF pathway activation. This explains our observation of absent SDHB activity in patient 5, who carries a VHL mutation, potentially mediated by HIF-1α-driven suppression of SDHB via miR-210 [30]. VHL-related PCCs are more likely to be identified at a younger age and detected incidentally by abdominal imaging in asymptomatic normotensive individuals [31].

Surprisingly, MN levels were highly elevated either isolated (patient 4) or predominantly (patient 5) in VHL carriers in our cohort, which is not usual for VHL syndrome. In contrast, patient 10 had bilateral PCCs with isolated NMN hypersecretion at a young age accompanied by a pancreatic tumor; this clinical phenotype is characteristic of VHL syndrome, but we did not discover a VHL mutation. Thus, we could suggest that correlations between phenotype and genotype in patients with PCC are not as strict as it was thought before.

Patient 12, carrying a loss-of-function mutation in NF1, showed increased expression of genes in the PI3K/AKT/mTOR pathways, consistent with NF1 pathogenesis. Patient 2, with co-occurring pathogenic germline mutations in RET and SDHB, exhibited the most distinct transcription profile. He had usual-for-MEN2 (type b) syndrome clinical signs (bilateral PCCs, medullar thyroid cancer) [32].

Notably, the profiles of patients 1, 2, and 5 differed significantly from both controls and other patients. Germline mutations were detected in patients 2, 5, and 12, but not in patient 1. Furthermore, immunohistochemistry revealed a lack of SDHB activity in patient 1, indicating dysfunction of the SDHx gene complex. The absence of a detectable germline mutation suggests the presence of a somatic mutation or an alternative inactivating mechanism, the exact nature of which remains to be identified.

A simultaneous increase in the transcription of Krebs cycle genes and PI3K/AKT/mTOR pathway genes was observed in patients 1, 2, 3, and 5. This coordinated shift is characteristic of proliferating cells, where the Krebs cycle intermediates are diverted for biosynthesis (cataplerosis), a process fueled and regulated by growth factor signaling and involving factors like c-Myc and HIF-1α [14]. The observed profiles may thus reflect a cellular state akin to early tumorigenic stress.

It is crucial to interpret transcriptional changes with caution, as they are not solely determined by genetics. Expression is a dynamic process influenced by numerous external and internal factors. For example, RET transcription can be modulated by interleukin IL-8 [33], NGFR by systemic inflammation [34], and MTOR by inflammatory cytokines like TNF-α [35]. Metabolic factors such as insulin, glucose, and glutamine levels can also significantly influence the PI3K/AKT/mTOR pathway and Krebs cycle gene expression [36], while reactive oxygen species can inhibit the cycle by affecting ACO2 [37].

This pilot study highlights the potential value of transcriptional analysis in the postoperative management of PCC for revealing underlying predispositions. However, the findings must be interpreted within the context of several limitations: the small sample size, imperfect control matching, the use of a limited set of reference genes for normalization, targeted genotyping of only four genes, and the exploratory nature of the dimensionality reduction analysis. Despite these constraints, this work represents a novel investigation into marker gene transcription in this specific clinical context and should be considered a pilot study requiring validation in a larger, independent cohort.

From a translational perspective, our findings, though preliminary, point to possible clinical utilities. First, blood transcriptomic profiling could aid in postoperative risk stratification, particularly for patients without a known germline mutation, by identifying those with a systemic molecular imbalance suggestive of a latent predisposition. Such patients might benefit from enhanced surveillance. Second, this noninvasive method could guide the application of more comprehensive genetic testing (e.g., whole-genome sequencing) by flagging individuals with a high likelihood of harboring cryptic regulatory variants. Finally, establishing a baseline transcriptional profile post-surgery opens up the possibility for longitudinal monitoring to detect early signs of recurrence. These potential applications, however, are contingent upon validation in larger, prospective cohorts, which should be the focus of future research.

5. Conclusions

We identified a minimal gene set capable of distinguishing certain patients after PCC resection from controls, suggesting a persistent, systemic dysregulation within the relevant gene networks (Krebs cycle, hypoxia, and kinase signaling pathways) even in the post-surgical state. These findings support the hypothesis that transcriptional profiling in the blood could serve as a functional biomarker for underlying genetic predisposition, potentially capturing alterations missed by standard coding sequence analysis. The integration of blood-based transcriptomics as a noninvasive screening tool could enhance postoperative risk stratification and guide the application of more comprehensive (and costly) genetic and epigenetic analyses. The results of this pilot study form the basis for a larger validation project in an expanded, multi-center cohort of post-surgical PPGL patients.