Introduction

The endocrine system includes several organs all devoted to the physiologic role of maintaining homeostasis and mediating medium- to long-term reactions of the human body to adapt it to external modifications. Tumors of the main endocrine glands, anterior pituitary (adenohypophysis), thyroid and parathyroid glands, adrenal cortex, adrenal medulla, and paraganglia are the object of this review. Tumors of the diffuse neuroendocrine system are not included, excellent reviews have comprehensively covered the topic [1]. A variety of tumors and nodules develop in endocrine glands, with different pathologic features and clinical behavior. Some are very common. Indeed, small indolent foci of papillary carcinoma are found in ~ 35% of well-sampled thyroid glands at autopsy. Others, like parathyroid carcinoma, are very aggressive, but also very rare. A significant minority of endocrine gland tumors develop in the context of inherited syndromes and paraganglionic tumors of the adrenal medulla and paraganglia have the highest degree of hereditability among human neoplasms. Table 1 summarizes inherited syndromes of endocrine tumors [2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18]. Our knowledge of the molecular pathogenesis of endocrine gland tumors has exploded in the recent past due to the application of high-throughput molecular analysis. These studies show a remarkable correlation between genotype and histologic phenotype. They are also allowing us to refine risk stratification for prognostic purposes, as well as providing targets for molecular therapy in the case of aggressive endocrine gland carcinomas. The purpose of this review is to summarize the principal findings and innovations in the field of endocrine gland tumors in order to provide a state-of-the-art outline of molecular alterations and their clinicopathologic relevance.

Table 1 Hereditary syndromes associated with endocrine tumors
Full size table

PitNET (pituitary adenoma): molecular pathology and correlation with clinicopathologic features

Pituitary adenomas, now termed pituitary neuroendocrine tumors (PitNET), originate from the six neuroendocrine hormone-secreting cell types derived from three main lineages: SF1-lineage gonadotrophs, TPIT-lineage corticotrophs, PIT1-lineage somatotrophs, lactotrophs, mammosomatotrophs, and thyrotrophs. Examples of PitNET are illustrated in Fig. 1 [2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25]. Table 2 is a summary of the main genes mutated in PitNET [19,20,21,22,23,24, 26].

Fig. 1
figure 1

PitNET (pituitary adenoma). GH-producing densely granulated PitNET (A, hematoxylin and eosin; B, GH immunohistochemistry): GNAS1 mutations occur in a subset of GH-producing pituitary adenoma, more frequently in densely granulated tumors. PRL-producing sparsely granulated PitNET with Golgi pattern PRL staining, the tumor was mitotically active and eventually metastasized to the brainstem and cerebellum (C, hematoxylin and eosin; D, PRL immunohistochemistry): the molecular pathogenesis of metastatic PitNET is still unclear

Full size image
Table 2 Somatic genetic alterations of PitNET (Pituitary adenoma) and their clinicopathologic relevance
Full size table

Although the majority of PitNET lacks known recurrent driver mutations, several — sometimes involving hormone synthesis pathways — have been identified in subsets of sporadic tumors. In addition, a small percentage of PitNET affects patients with inherited predisposition due to germline genetic alterations (Table 1). In some instances, these alterations can be inferred by immunohistochemistry but need to be confirmed by sequence analysis. An example is the immunohistochemical loss of Menin in the tumors of patients with MEN1 syndrome [38,39,40]. To date, morphologic classification, together with clinical and radiologic data, continues to be the best predictor of patient prognosis and therapy response. The relevance of genetic profiling for the diagnostic process is still being defined [19,20,21,22,23,24,25]. Epigenetic alterations, particularly those connected to chromatin remodeling and cell cycle regulation [25], play a major role in the development of PitNET, independently of the hormone-producing phenotype of the tumor. Genes frequently dysregulated by epigenetic modifications include the following: CDKN2A, RB1, DAPK1, GADD45G, THBS1, RASSF1A, FGFR2, MGMT, CASP8, TP73, HMGA1, HMGA2 [41]. In general, chromosomal abnormalities do not correlate with prognosis but are more common in hormone-producing tumors compared with those not associated with hormone production (silent PitNET) [23].

Activating GNAS mutations have been reported in 40–60% of sporadic densely granulated somatotroph PitNET and are present as a mosaic in the 10–15% of patients with McCune–Albright syndrome that have excess GH, usually due to GH-producing cell hyperplasia and less commonly to a GH-producing PitNET [42]. Mammosomatotrope PitNET also exhibits GNAS-related alterations [23]. Conversely, in sparsely granulated somatotroph tumors, somatic mutations of the GH receptor (GHR) altering GH autoregulation and STAT signaling have been reported [20, 25].

In corticotroph PitNET, ATRX mutations correlate with aggressive biological behavior and distant metastasis [43]. Densely granulated biochemically functioning corticotroph tumors harbor USP8 [44], USP48, and less frequently BRAF p.V600E mutations [22]. The role of these changes and their potential therapeutic implications are still controversial [24].

The distinctive molecular signature of lactotroph PitNET includes epigenomic alterations such as high expression of MYC targets and dopamine receptor D2 (DRD2) [23]. However, the SF3B1 p.R625H hotspot mutation has been recently discovered in some lactotroph tumors characterized by high prolactin levels and short progression-free survival [45]. Furthermore, somatic SDHA mutations and SDHD loss of heterozygosity have been reported in rare spontaneous PRL-producing macrotumors [16, 46].

The molecular pathogenesis of metastatic PitNETs is still unclear, due to the rarity of these tumors. ATRX [19, 47] and PTEN [43] mutations have all been reported in some metastatic PitNETs.

Thyroid tumors: molecular pathology and correlation with clinicopathologic features

Tumors of the thyroid gland enjoy a remarkable correlation between histologic phenotype and genotype. This correlation has contributed to refining the current classification scheme. The vast majority of tumors arising in the thyroid are of follicular cell derivation, most are benign, and when endowed with malignant potential, usually follow a very favorable clinical course. This generally favorable course is due to the first effective form of a molecularly targeted therapy, radioactive iodide treatment [48]. A small proportion of tumors are neuroendocrine, originating from parafollicular cells (C-cells). Since they always have malignant potential, they are classified as medullary carcinoma, which represents ~ 3–5% of all carcinomas of the thyroid gland. Up to 25% of medullary carcinoma is inherited in the context of MEN syndromes (Table 1).

Based on clinical outcome, malignant tumors of follicular cells are broadly divided into three groups: those that have a favorable prognosis, anaplastic (undifferentiated) thyroid carcinoma characterized by a very poor prognosis, and a third group of tumors that have intermediate prognosis. While tumors in the first group are histologically well differentiated with clearly defined papillary or follicular architecture or are composed of clearly recognizable oncocytic cells, tumors with very poor prognosis are undifferentiated (i.e., anaplastic). Tumors in the group with intermediate prognosis are often poorly differentiated but may also retain conventional histologic differentiation (papillary, follicular, oncocytic). Under the microscope, they have in common with the prognostically favorable tumor group at least some degree of histologic differentiation, while they share with anaplastic carcinoma high-grade features, i.e., the presence of high mitotic activity and/or tumor necrosis. This classification scheme for thyroid carcinoma of follicular cells based on prognosis is clinically relevant and has been endorsed by the latest 5th edition of the World Health Organization (WHO) scheme (Table 3). The group of tumors that are well differentiated is in turn histologically divided into three subgroups. The first subgroup is composed of tumors that are follicular patterned, which include follicular adenoma and follicular carcinoma (follicular carcinoma when there is the invasion of tumor capsule or of blood vessels), as well as tumors of the encapsulated follicular variant papillary carcinoma family: encapsulated follicular variant papillary carcinoma when there is the invasion of tumor capsule or of blood vessels, and NIFTP (non-invasive follicular thyroid neoplasm with papillary-like nuclear features) when no invasion can be identified [49]. These tumors have a RAS-like molecular signature following the 2014 TCGA molecular classification scheme [50] as discussed in the next paragraph. The second subgroup is that of conventional (i.e., not encapsulated follicular variant type) papillary carcinoma, characterized by the well-known alterations of nuclear morphology (nuclear clearing, irregular contours of the nuclear membrane, grooves, and pseudoinclusions) [49]. These tumors are characterized by infiltrative growth and typically make papillae, although sometimes they can have less typical features, such as infiltrative follicular or solid/trabecular growth, or other less common features that characterize the numerous papillary carcinoma subtypes [51]. These tumors have a BRAF p.V600E-like molecular signature following the 2014 TCGA molecular classification scheme [50], as discussed in the next paragraph. The third subgroup is that in which tumor cells are oncocytic and lack the nuclear alterations of papillary carcinoma. These tumors are characterized by homoplasmic mtDNA mutations [52] associated with dramatic DNA copy-number alterations with widespread loss of heterozygosity [53], as discussed in the next paragraph.

Table 3 Prognostically relevant classification of tumors of follicular cell origin
Full size table

Medullary thyroid carcinoma is the primary neuroendocrine tumor of the thyroid gland. In spite of the remarkable variability of cell morphology and growth patterns (none of which is prognostically relevant), the only subtype recognized by the current WHO 5th edition is the medullary microcarcinoma, i.e., a tumor measuring less than 10 mm (or less than 5 mm according to some authors) scheme. The WHO 5th edition emphasizes the importance of proliferative grading for medullary carcinoma, following the International Medullary Thyroid Carcinoma Grading System (IMTCGS) [54]. The IMTCGS, based on the evaluation of mitotic count and tumor necrosis (Table 4), is in line with the classification framework of neuroendocrine neoplasms [55].

Table 4 International Medullary Thyroid Carcinoma Grading System (IMTCGS)
Full size table

The molecular landscape of thyroid tumors, particularly that of follicular cell derivation, has come into focus also thanks to next-generation sequencing and other high-throughput methods. One of the forces driving these studies has been the need to identify genomic alterations that can be targeted by pathway-specific molecular drugs in aggressive carcinomas that do not respond to conventional radioiodine therapy [54, 56,57,58]. Table 5 is a summary of the main genes involved in thyroid tumor development and progression and of their clinicopathologic relevance. Overall, results are very consistent and converge on several important points:

Table 5 Genetic alterations of thyroid tumors and their clinicopathologic relevance
Full size table
  1. i.

    Genetic alterations include “Early/Driver” molecular changes and “Late/Progression associated” events [28, 50, 62, 63, 79]. These are illustrated in Figs. 2 and 3. Examples of tumors with “Early/Driver” alterations are shown in Fig. 4.

    Fig. 2
    figure 2

    Modified from: Acquaviva G. et al. [28]

    Genetic alterations of thyroid tumors. Genetic alterations include “Early/Driver” molecular changes (BRAF p.V600E-like for conventional papillary carcinoma, RAS-like for follicular patterned tumors, and coexistence of mtDNA mutations with severe DNA copy-number alterations for oncocytic tumors) as well as “Late/Progression-associated” molecular changes such as TERT promoter mutation, TP53 mutation, alterations of the PI3K/PTEN/AKT pathway in high-grade non-anaplastic carcinoma of follicular cells, and anaplastic thyroid carcinoma. PDTC, poorly differentiated thyroid carcinoma; DHGTC, differentiated high-grade non-anaplastic thyroid carcinoma; GH-CNV, genome haploidization-type DNA copy number variation leading to copy number neutral uniparental disomy.

    Full size image
    Fig. 3
    figure 3

    Modified from: Volante et al. [79]

    Driver molecular alterations, tumor type, and progression in thyroid tumors of follicular cells. PTC, papillary thyroid carcinoma; E-FVPTC, encapsulated follicular variant papillary thyroid carcinoma; PDTC, poorly differentiated thyroid carcinoma; GH-CNV, genome haploidization-type DNA copy number variation leading to uniparental disomy/aneuploidy.

    Full size image
    Fig. 4
    figure 4

    Thyroid tumors. NIFTP (A, hematoxylin and eosin): follicular patterned tumors such as NIFTP have RAS-like molecular signature. Papillary carcinoma (B, hematoxylin and eosin): conventional papillary carcinoma has BRAF p.V600E-like molecular signature, this case featuring glomeruloid papillae harbors a TPR::NTRK1 rearrangement. Oncocytic carcinoma (C, hematoxylin and eosin): oncocytic tumors have both mtDNA mutations and severe DNA copy number alterations. High-grade non-anaplastic papillary carcinoma with tumor necrosis (D, hematoxylin and eosin), poorly differentiated thyroid carcinoma (E, hematoxylin and eosin), and anaplastic thyroid carcinoma (F, hematoxylin and eosin): these aggressive high-grade tumors harbor early/driver molecular alterations and additional mutations associated with tumor progression affecting TERT promoter and PI3K/PTEN/AKT pathway genes; inactivating TP53 mutations are typically associated with anaplastic carcinoma

    Full size image
  2. ii.

    “Early/Driver” alterations are mutually exclusive [28, 50, 62, 63, 79]. They are commonly used for molecular analysis of preoperative fine needle aspiration specimens. The features as well as the pros and cons of the starting material for this type of analysis are summarized in Table 6.

    Table 6 Starting material for preoperative molecular analysis of thyroid nodules
    Full size table
  3. iii.

    “Late/Progression associated” alterations are found in combination with “Early/Driver” changes, consistent with a general model of multi-step progression from well-differentiated to undifferentiated carcinoma. In cases where poorly or undifferentiated areas are associated with a well-differentiated component, “Early/Driver” alterations are identified in both areas, while “Late/Progression associated” changes are restricted to the less differentiated portions of the tumor [80]. Thus, the number of mutations per tumor increases from well-differentiated to undifferentiated carcinoma. Mutation burden is highest in anaplastic carcinoma, lowest in conventional papillary carcinoma, and intermediate in aggressive/advanced papillary and follicular carcinoma [81]. Examples of tumors with “Late/Progression associated” alterations are shown in Fig. 4.

  4. iv.

    RAS mutations or equivalent molecular alteration (RAS-like tumors) are “Early/Driver” events (see paragraph (ii)) for follicular patterned tumors (Figs. 2, 3, and 4 and Table 5). RAS-like tumors have a homogeneous molecular profile, low MAPK-signaling (because of ERK to RAF monomer negative feedback), high differentiation score, and are malignant only if there is an invasion of the tumor capsule or blood vessels [50].

  5. v.

    BRAF p.V600E mutation or equivalent molecular alterations (BRAF p.V600E-like tumors) are “Early/Driver” events (see paragraph (ii)) for conventional papillary carcinoma (Figs. 2, 3, and 4 and Table 5). BRAF p.V600E-like tumors have a heterogeneous molecular profile, high MAPK-signaling (because of the lack of ERK to RAF monomer negative feedback), and low differentiation score (based on the level of expression of 16 thyroid metabolism and function genes, e.g., TG, TPO, PAX8).

  6. vi.

    Coexistence of mtDNA mutations with severe DNA copy-number alterations represents the “Early/Driver” event (see paragraph (ii)) for oncocytic tumors (Figs. 2, 3, and 4 and Table 5). Mutations of mitochondrial DNA (mtDNA) are homoplasmic mtDNA and mostly affect mitochondrial genes encoding Complex I of the respiratory chain [52]. DNA copy-number alterations are dramatic, with widespread loss of heterozygosity and loss of chromosomal DNA, following genome-wide DNA haploidization and copy-number-neutral uniparental disomy [53, 82, 83]. This pathway is unique to oncocytic tumors, which typically do not carry conventional BRAF-like or RAS-like alterations [82,83,84]. While mtDNA mutations are responsible for the oncocytic phenotype [52], the loss of chromosomal DNA is linked to tumor development, since genome haploidization-type DNA copy-number changes are more common in cases diagnosed histologically as oncocytic carcinoma as opposed to oncocytic adenomas and are rare in hyperplastic oncocytic nodules. This has potentially important implications for molecular testing of preoperative fine needle aspiration, since conventional BRAF-like or RAS-like alterations are commonly absent in oncocytic tumors, while Bethesda category III and IV with oncocytic changes have a higher prevalence of DNA copy-number alterations compared with the same categories without cytologically identified oncocytic morphology [85].

  7. vii.

    “Late/Progression-associated” alterations include mostly somatic mutations of TP53, TERT promoter, and dysregulation of the PI3K/PTEN/AKT pathway (Figs. 2 and 3 and Table 5). Mutations of CDKN2A, of SWI/SNF (switch/sucrose non-fermentable) chromatin remodeling complex genes (ARID1A, ARID1B, ARID2, ARID5B, SMARCB1, PBRM1, ATRX), of Histone methyltransferase genes (KMT2A, KMT2C, KMT2D, SETD2), and of DNA mismatch repair (MMR) genes (MSH2, MSH6, and MLH1) have also been reported [56, 62, 63]. “Late/Progression-associated” changes, in particular TERT promoter mutations, can be utilized for risk stratification, also using preoperative fine needle aspiration specimens [57].

  8. viii.

    TERT promoter mutations are more frequent and have higher mutated allelic fraction in poorly differentiated, anaplastic, and aggressive/advanced cancers (including high-grade papillary carcinoma) compared with well-differentiated carcinoma [56, 62, 63].

  9. ix.

    TERT promoter and particularly its co-mutation with BRAF p.V600E or RAS is a powerful marker of poor outcome. Aggressive/advanced papillary carcinomas, many of which are histologically high-grade, have at least one of three genetic alterations: duplication of chromosome 1q, duplication of chromosome 5 p harboring the TERT genomic locus, and TERT promoter mutation (THYT1 signature) [57].

  10. x.

    TP53 mutation has the highest prevalence in anaplastic carcinoma compared to all forms of advanced/aggressive thyroid carcinoma, including both poorly differentiated and high-grade papillary carcinoma [56, 62, 63, 66].

  11. xi.

    Rearrangements — such as RET/PTC, NTRK1, NTRK3, and PAX8-PPRG not rare in well-differentiated tumors — are uncommon [56, 62, 63].

Parathyroid tumors: molecular pathology and correlation with clinicopathologic features

The spectrum of parathyroid tumors includes adenoma, atypical tumor (neoplasm of uncertain malignant potential, previously defined as “atypical parathyroid adenoma”), and carcinoma. These tumors can arise in any gland including ectopic ones or areas where embryonic parathyroid remnants may be found [86]. The majority of them cause primary hyperparathyroidism, with adenomas accounting for at least 85% of cases [87, 88].

In general, immunohistochemistry is not necessary for diagnostic purposes. However, it may be useful to screen for hereditary conditions associated with inactivation of genes such as CDC73 and MEN1, causing hyperparathyroidism-jaw tumor (HPTJT) and MEN1 syndrome, respectively [89]. These hereditary conditions usually present with multiglandular involvement and/or multinodular tumors. Parathyroid tumors that arise in the context of hyperparathyroidism-jaw tumor (HPTJT) syndrome feature eosinophilic cytoplasm, perinuclear clearing, nuclear expansion, micro-cystic structures, and sheet-like growth pattern [90].

The main genes mutated in parathyroid tumors are summarized in Table 7 and illustrated in Fig. 5. Examples of parathyroid tumors are shown in Fig. 6. The molecular pathogenesis underlying the majority of sporadic parathyroid adenoma remains unknown. Syndromic parathyroid adenomas constitute ~ 10% of cases [91], found in MEN1, MEN2, MEN4, HPTJT, and isolated familial hyperparathyroidism (FIHP). These syndromes have recently been complemented by MEN5, associated with hereditary mutations of MAX [92] (Table 1). In these inherited conditions, the parathyroid glands contain multiple clonal adenomas which mimic the clinical appearance of cases traditionally diagnosed as parathyroid hyperplasia [93, 94].

Table 7 Somatic genetic alterations of parathyroid tumors and their clinicopathologic relevance
Full size table
Fig. 5
figure 5

Genetic alterations of parathyroid tumors. MEN1 loss of function represents the most common alteration in parathyroid adenoma. The most common alteration of parathyroid carcinoma is CDC73 inactivation, also found in a minority of atypical parathyroid tumors

Full size image
Fig. 6
figure 6

Parathyroid carcinoma. There is mitotic activity, nuclear pleomorphism (A, hematoxylin and eosin), and parafibromin expression is lost (B, parafibromin immunohistochemistry)

Full size image

In sporadic adenoma, the most common somatic alteration is inactivation of MEN1. This is caused by loss of heterozygosity (LOH) due to large deletions or genetic recombination at 11q13 (where MEN1 is located) found in ~ 35% of all parathyroid adenoma and/or MEN1 inactivating mutations found in up to ~ 20% of cases [105]. In addition to LOH and somatic mutations, other mechanisms can lead to the inactivation of MEN1, including epigenetic silencing. Interestingly, biallelic MEN1 inactivation occurs in approximately 50% of cases in which LOH at 11q13 is detected, raising the hypothesis that other genes on 11q may also play a role in tumor development.

Cyclin D1 (also located at 11q13) is overexpressed in 10–40% of parathyroid adenomas due to aberrant promoter methylation of different cyclin-dependent kinase inhibitors (CDKIs), while rearrangement of the Cyclin D1 gene (CCND1) occurs in up to ~ 10% of parathyroid adenomas [105].

Other somatic mutations involving CDKN1B (encoding p27), EZH2 (encoding the zinc-finger protein X-linked transcription factor), ASXL3, and MTOR have been reported in a small minority of parathyroid adenomas [105,106,107]. Somatic CDC73 mutations are rare in adenomas. They have been reported only in atypical parathyroid tumors, in some adenomas in the context of HPTJT, and in a small number of cystic adenomas [108]. Interestingly, parathyroid nodules in secondary or tertiary hyperparathyroidism — typically associated with chronic renal failure — harbor different somatic changes compared with those of adenomas in primary hyperparathyroidism [106].

Parathyroid carcinoma is rare and the majority of cases are sporadic. Parathyroid carcinoma develops in 10–15% of patients with HPJT and FIHP (Table 1), while it is uncommon in other inherited conditions [90, 109, 110]. Only a few studies have been conducted on the molecular pathogenesis of sporadic carcinomas. Contrary to parathyroid adenomas, parathyroid cancer rarely exhibits MEN1 mutations [111]. Inactivating CDC73 alterations have been reported in 40–80% of sporadic cases [111,112,113]. CDC73 alterations include truncating or frameshift mutations, as well as missense mutations leading to the loss of parafibromin immunoreactivity [114]. CDC73-mutant parathyroid carcinomas exhibit higher genomic instability with DNA copy number changes, greater mutational burden, and worse patient outcomes compared with wild-type cases [115].

Loss of TP53 and RB1 alleles, CCND1 (encoding Cyclin D1) amplification, and TERT promoter mutations have been reported [116, 117]. PTEN, NF1, KDR, and PIK3CA mutations may represent potential targets for molecular therapy [118]. Metastatic parathyroid carcinoma has a different expression profile compared with non-metastatic parathyroid carcinoma and parathyroid adenoma [119, 120]. Several epigenetic alterations have been discovered in parathyroid carcinomas, including aberrant methylation of APC and of the cell cycle regulators CDKN2A and CDKN2B [121].

Adrenal cortical tumors: molecular pathology and correlation with clinicopathologic features

The spectrum of endocrine tumors of the adrenal cortex includes adrenocortical nodular disease, adrenal cortical adenoma, and adrenal cortical carcinoma. Recently, molecular insights have led to modify the terminology related to adrenocortical nodular disease [122] which currently includes several types of clonal benign proliferations: sporadic nodular adrenocortical disease (a common condition), micronodular adrenocortical disease (a rare condition), and bilateral macronodular adrenocortical disease (a rare condition). Micronodular and bilateral macronodular adrenocortical diseases are often associated with germline pathogenic mutations of several genes [123,124,125]. Our understanding of genomic and hormonal landscapes of adrenal cortical adenoma has also advanced significantly [126], and genotype–phenotype correlations have been proposed for both aldosterone-producing [127, 128] and cortisol-secreting adenomas [129]. Importantly, it is not uncommon for a single alteration to affect different functional pathways, as has been demonstrated for KCNJ5 mutations [130]. Concerning the pathogenesis of cortical carcinoma, several main pathways of tumorigenesis have been discovered, involving cell cycle regulation, Wnt signaling, chromosome maintenance/chromatin remodeling, and the PKA pathway [131] (Fig. 7).

Fig. 7
figure 7

Genetic alterations of adrenal cortical tumors. Adenomas arise as a result of mutations affecting two main groups of genes: the aldosterone-producing adenomas harbor most frequently mutations for KCNJ5 or the ion channel encoding genes, while cortisol-producing adenomas often develop due to alterations in the PKA pathway, typically PRKACA. Genetic alterations of carcinomas mainly involve TP53 but also genes commonly mutated in non-functioning adenomas

Full size image

Furthermore, integrated analysis of these findings with transcriptomic data, epigenetic findings, and copy number changes has led to the identification of three main classes of adrenal cortex carcinoma, with important clinical and prognostic implications [132,133,134].

Adrenocortical nodular disease

Nodular adrenocortical disease with bilateral involvement of the adrenal cortex rarely occurs in young patients, but when present, it is frequently associated with germline conditions (Table 1). Germline variants of PRKAR1A, PRKACA, PDE11A, PDE8B, and 2p16 CNC2 locus alterations are frequently reported in micronodular adrenocortical disease with bilateral involvement of adrenal cortex, which typically affects children and young adults [135] (Table 1). Germline PRKAR1A mutations (less frequently of PDE8B and PDE11A) cause Primary Pigmented Nodular Adrenocortical Disease, a distinct subtype of bilateral micronodular adrenocortical disease typically found in association with Carney’s complex [136]. The bilateral macronodular adrenocortical disease is caused by pathogenic ARMC5 variants (~ 50% of cases). Further alterations may involve the following genes: MEN1, FH (Hereditary Leiomyomatosis and Renal Cell Cancer), APC (Familial Adenomatosis Polyposis), GNAS (McCune Albright Syndrome), and the rarely mutated PDE11A, PDE8B, and 2p16 CNC2 locus [123, 125, 132, 136, 137]. Variable patterns of Cushing syndrome are typical clinical manifestations of both micro- and macronodular adrenocortical disease.

Adrenal cortical adenoma

Cortical adenomas are the most common tumors of the adrenal cortex.

Alterations of distinctive pathways (active under normal conditions) involved in the physiologic production of aldosterone and cortisol are typical of the corresponding hormone-producing adenoma. Indeed, functioning adenomas that cause primary aldosteronism harbor specific somatic mutations of several ion channel genes which lead to both cellular proliferation and increased aldosterone production in the cells of the zona glomerulosa [126]. They are mutually exclusive and involve KCNJ5 (K + channel) [138], ATP1A1 (Na + /K + channel) [139], ATP2B3 (Ca2 + channel) [139], CACNA1D (Ca2 + channel), CACNA1H (Ca2 + channel) [140], and CLCN2 (Cl- channel) [95, 141]. KCNJ5 mutated adenomas account for the large majority of the cases (~ 40% of aldosterone-producing adenomas) and tend to mainly affect young female patients [137, 142]. Recurrent phenotypic and clinical characteristics have been identified in ion channel gene mutated adenomas [127, 128]. As shown in Table 8, these include the expression of steroidogenic enzymes, cytomorphology, and lateralization index of the adrenal vein sampling (AVS). Interestingly, ion channel genes such as KCNJ5, CACNA1H, CACNA1D, or CLCN2 may also be mutated in the germline, causing familial aldosteronism [126]. Somatic mutations of CTNNB1 — encoding the Wnt-pathway effector beta-catenin — occur in ~ 5% of sporadic aldosterone-producing adenomas and have been correlated to delayed disease onset and female prevalence [96].

Table 8 Genetic alterations of adrenal cortical tumors and their clinicopathologic relevance
Full size table

Functioning adenomas that produce cortisol feature genetic alterations of the protein kinase A (PKA) pathway active under normal conditions in the production of cortisol. The PKA pathway is physiologically activated by ACTH so that PKA catalytic subunits (PKA-C) can enter the nucleus of zona fasciculata cells enhancing transcription of genes that promote cell proliferation and synthesis of cortisol [126]. PKA pathway genetic alterations of cortisol-producing adenomas affect most frequently the following genes: PRKACA (~ 40% of cases), PRKAR1A, GNAS, or PRKACB [97, 98]. As for sexual steroid-producing adenomas, the molecular pathogenesis remains largely unknown. CTNNB1 mutations are the most frequent molecular alterations of cortical adenomas not associated with hormone production (silent adenomas), particularly in large-sized tumors [129].

Adrenal cortical carcinoma

Adrenal cortical carcinoma is rare and mostly associated with somatic genetic changes (Figs. 7 and 8). Common clinical presentations include Cushing or adrenogenital (virilization-feminization) syndromes due to hormone production [99, 122]. Hereditary cases typically affect children, with up to 80% of pediatric cases carrying germline mutations. The most common mutation affects TP53 (Li-fraumeni syndrome), followed by alterations of the mismatch repair system (Lynch syndrome) [100, 101]. Additional hereditary conditions include Beckwith-Wideman syndrome and MEN1 [100]. Genes frequently altered in benign conditions such as PRKAR1A, MSH2, APC, MEN1, and NF1 are mutated only in small subsets of carcinomas [132]. The most common genetic signatures affect the cell cycle, Wnt signaling, and chromatin remodeling. TP53 mutations found in ~ 20% of adrenal cortical carcinomas are the most common changes [143]. Recurrent somatic genetic alterations affect other cell cycle regulatory genes such as RB1, CDK2NA [132], MDM2, and CDK4 [131, 133]. Wnt signaling is dysregulated by CTNNB1 mutations, ZNRF3 mutations, and deletions [132, 143]. Importantly, TP53 or Wnt pathway mutations are typically mutually exclusive but similarly associated with poor prognosis [132, 133, 143]. Dysregulation of chromatin remodeling is caused by the alteration of several genes, such as ATRX and DAXX [132].

Fig. 8
figure 8

Adrenal cortical tumors. Adrenal cortical adenoma composed by lipid-rich cells resembling the zona fasciculata with low mitotic activity (A, hematoxylin and eosin): cortisol-producing adenoma such as the one shown in the picture has PKA pathway alterations. Adrenal cortical carcinoma with nuclear pleomorphism, mitotic activity, and trabecular growth pattern (B, hematoxylin and eosin): TP53 is the gene most commonly mutated

Full size image

Furthermore, structural alterations (rearrangements and deletions) in tumor DNA at 11p15 are a frequent finding and can cause loss of imprinted H19 tumor suppressor gene and overexpression of the IGF2 oncogene [103]. IGF2 overexpression in adrenal cortical carcinoma can be identified by immunohistochemistry and may be useful in the differential diagnosis with adenoma [104].

Chromosomal alterations are heterogeneous and have been clustered into three groups: those with extensive chromosome loss (~ 50%), those with variable levels of ploidy (~ 40% — the group of chromosomal changes with worse prognosis), and those with limited chromosomal DNA alteration (~ 10%).

Comprehensive molecular classification of adrenal cortical carcinoma is evolving [126, 143]. Over the years, evidence provided by mutation analysis, chromosomal, methylation, and transcriptome profiling has been integrated to define prognostic groups for risk stratification [126, 133, 134, 144]. According to an important study by Assié et al., there are two main types of adrenal cortical carcinoma [132]. The set of “CpG island methylator phenotype-low” (CIMP-low) carcinomas has infrequent alterations in TP53 or Wnt pathways, mRNA expression pattern predictive of less severe prognosis, chromosome loss, and low rate of disease progression. “CIMP-high” carcinomas typically have alterations of TP53 or Wnt pathway, mRNA expression pattern predictive of poor prognosis, whole-genome duplication, and high rate of disease progression. The TCGA has built on this experience and has proposed an integrated molecular classification model based on DNA copy number, DNA methylation, mRNA expression, and miRNA expression profiles [132,133,134]. This classification model has three groups — termed Cluster-1, -2, -3 — which have been defined after Cluster of Cluster (CoC) analysis. Each CoC cluster is highly relevant for patient prognosis, with Cluster 3 being associated with worse outcome [132,133,134].

Paraganglionic tumors (tumors of the adrenal medulla and of extra-adrenal paraganglia): molecular pathology and correlation with clinicopathologic features

The main genes mutated in paraganglionic tumors are illustrated in Fig. 9. An example of a paraganglionic tumor is shown in Fig. 10.

Fig. 9
figure 9

Genetic alterations of paraganglionic tumors. Genetic alterations in PPGL have been grouped into three main groups reflecting three different mechanisms of tumorigenesis: the Cluster 1-pseudohypoxia pathway, characterized by genetic alterations of the HIF1-alpha-activated response to hypoxia pathway and of other similar hypoxia-inducible factor gene pathways; the Cluster 2-Kinase signaling, including the most common alterations identified in paraganglionic tumors such as NF1 mutations; the more recently described Cluster 3-Wnt/Sonic Hedgehog pathway. *Germline and somatic alterations. **Only somatic alterations

Full size image
Fig. 10
figure 10

Paraganglionic tumors. Cells resembling normal chromaffin cells with abundant granular cytoplasm are arranged in well-defined nests (A, hematoxylin and eosin). Head and neck paraganglioma, such as the one shown in the picture, often has inactivating SDHD mutations. If any of the subunits of the SDH complex is altered, the entire complex becomes unstable and immunohistochemical SDHB expression is lost (B, SDHB immunohistochemistry)

Full size image

Paraganglionic tumors are neuroendocrine neoplasms that develop from neural crest-derived progenitors in the adrenal medulla (Pheochromocytoma) and paraganglia (Paraganglioma), respectively. In the latest WHO classification of endocrine tumors, they are considered malignant, although the overall proportion of cases that metastasize is low, ~ 10% of cases. Paragangliomas are further classified in sympathetic and parasympathetic neoplasms according to cell origin and localization. In particular, sympathetic paraganglioma arises within sympathetic nerve plexuses, fibers, and pre- and paravertebral sympathetic chains — thus abdominal cavity, retroperitoneum, pelvis, and thorax are the most prevalent sites. The majority of parasympathetic paraganglioma develops from parasympathetic glomera, and tumors are typically found in the head and neck region, while pheochromocytoma develops in the adrenal medulla. Paraganglionic tumors share the same embryonic origin and are frequently associated with inherited germline mutations (Table 1). Indeed, paraganglionic tumors have the highest degree of hereditability among human neoplasms, with germline mutations in up to ~ 40–80% of cases (vs. ~ 10% in other tumor types) [145,146,147]. Most cases of sympathetic paraganglionic tumors are functional due to catecholamine production, causing hypertension in the majority of patients [148], while parasympathetic paraganglioma typically presents as asymptomatic masses. Sympathetic paraganglionic tumors in children account for up to 20% of cases [149, 150] and most are associated with germline mutations [102, 150]. Conversely, parasympathetic paragangliomas rarely affect children [102, 150, 151]. Germline mutations are an important risk factor for the development of all types of paraganglionic tumors, occurring in up to ~ 40 of adult cases and in up to 80% of pediatric ones. Patients with germline mutation often develop synchronous or metachronous multicentric tumors [102, 150, 152, 153].

Germline mutations mainly affect the following genes: RET, NF1, VHL, TMEM127, SDHA, SDHB, SDHC, SDHD, SDHAF2, FH, MAX, EPAS1, DLST, MDH2, GOT2, and DNMT3A (Table 1).

SDH (succinate dehydrogenase, complex II of the mitochondrial respiratory chain) consists of four subunits (SDHA, SDHB, SDHC, SDHD), and the genes encoding SDHB and SDHD are those more frequently mutated in the germline [153]. SDHD mutations cause the majority of paragangliomas of the head and neck region and typically present with single or multifocal tumors exclusively located in the head and neck. On the other hand, thoracoabdominal tumors more frequently harbor SDHB mutations. Paraganglionic tumors can also be found in association with gastrointestinal stromal tumors (GIST) and pulmonary chondromas in the so-called Carney triad, a nonhereditary condition characterized by epigenetic alterations of SDHC [154, 155]. Interestingly, somatic NF1, RET, and VHL are also found in sporadic tumors, with NF1 representing the gene most commonly mutated ( ~ 20% of cases) [156, 157]. Additional somatic mutations, not found in the germline, affect HRAS, BRAF, SETD2, FGFR1, TP53, ATRX, ARNT, IDH1, H3F3A, MET, and CSDE1 [146].

Paraganglionic tumors have been divided into three molecular clusters that are also recognized by the TCGA [145]. Cluster 1 tumors have a response to hypoxia pathway dysregulation (pseudohypoxia) characterized by increased transcription of genes targeted by hypoxia-inducible factors (HIF1-alpha and other factors) which promote angiogenesis, cell proliferation, survival, and epithelial-mesenchymal transition [158]. Of note, in SDH-/FH-deficient and mutant IDH tumors, oncometabolite accumulation induces DNA hypermethylation and other epigenetic changes. Cluster 2 tumors feature abnormal activation of RAS/RAF/ERK, PI3K/PTEN/AKT, and MYC/MAX/MXD1 pathways. They also exhibit a hypomethylated phenotype and frequent somatic copy number changes [146]. Cluster 3 tumors are characterized by dysregulation of Wnt and Sonic Hedgehog pathways. Indeed, sporadic MAML3 fusions and CSDE1 mutations in paraganglionic tumors that activate Wnt and Sonic Hedgehog pathways have been discovered to be major driving factors in tumor development [145]. Given the high hereditability of paraganglionic tumors, genetic screening is recommended, particularly in pediatric patients [150, 159]. In this respect, immunohistochemistry is a very useful screening test: if any of the subunits of the SDH complex is lost due to mutations or epigenetic alterations, the entire complex becomes unstable and the SDHB subunit is degraded in the cytoplasm. Loss of the SDHB protein can be demonstrated by SDHB immunohistochemistry, pointing to the need for SDH sequencing to confirm SDH subunit germline mutation [38]. This type of so-called “molecular-immunohistochemistry” can be applied not only to anticipate the genetic background of individual paraganglioma tumors but also to prevent erroneous diagnostic conclusions in the case of multiple lesions mimicking metastatic disease [146, 147].