Article Text

Download PDFPDF
Pathological and molecular features of adrenocortical carcinoma: an update
  1. M Volante1,
  2. C Buttigliero1,
  3. E Greco1,
  4. A Berruti2,
  5. M Papotti1
  1. 1
    Division of Pathology, Department of Clinical & Biological Sciences, University of Turin at San Luigi Hospital, Orbassano, Torino, Italy
  2. 2
    Division of Oncology, Department of Clinical & Biological Sciences, University of Turin at San Luigi Hospital, Orbassano, Torino, Italy
  1. Dr M Papotti, Department of Clinical and Biological Sciences, University of Turin at San Luigi Hospital, Regione Gonzole 10, I-10043 Orbassano, Torino, Italy; mauro.papotti{at}


The pathological diagnosis of adrenocortical carcinoma (ACC), which is based on gross and microscopic criteria, is subjective. None of the features are absolutely indicative of malignancy, although their combination in a scoring system may correctly identify ACC. The Weiss system, which is currently the most popular, combines nine morphological parameters, of which three are structural (“dark” cytoplasm, diffuse architecture, necrosis), three are cytological (atypia, mitotic count, atypical mitotic figures) and three are related to invasion (of sinusoids, veins and tumour capsule). Although there are strictly defined criteria for each feature, some are straightforward and objective, while others are potentially more problematic (diffuse architecture, necrosis, sinusoidal, venous and capsular invasions). The classification of oncocytic and paediatric adrenocortical tumours is even more challenging, as not all of the above morphological parameters are predictors of malignancy in these tumour types. As an alternative to the morphological approach, a wide array of chromosomal, genetic, molecular and immunohistochemical markers have been tested in ACC to identify reliable diagnostic and prognostic factors. Genetic and epigenetic alterations of p53, IGF-2 and molecules involved in cancer cell invasive properties seem the most promising. These molecular markers may not only play a role in the biology of these tumours and have prognostic implications, but may also be used as potential targets for treatment. However, these markers are not sufficiently sensitive and specific to replace conventional morphological criteria.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Adrenocortical carcinoma (ACC) is a rare tumour of the adrenal cortex with an incidence of approximately 1 per million, affecting women more frequently than men at any age. It has a bimodal age distribution, with one peak in early childhood and another in adults over age 60. The presence of distant metastases is a relatively common occurrence at diagnosis; tumour progression is observed in most cases, with a mean survival of less than three years.13 Combined treatment modalities, particularly mitotane-based therapy, have improved survival.4

The pathological diagnosis of ACC is difficult, due to the lack of clear-cut morphological criteria for malignant tumours.5 6 To overcome the difficulties of the standard microscopic evaluation, several studies have identified molecular factors having a diagnostic (and/or prognostic) significance. Data from gene array, cytogenetics, cell cycle analysis, receptor and growth factor expression, and expression of invasion/metastasis modulators have contributed to better defining the malignant phenotype of adrenocortical tumours, although no marker has so far been shown to be specific and sensitive enough for a diagnosis of ACC.

This review briefly summarises the traditional morphological criteria for ACC diagnosis and the relevant molecular markers differentially expressed in benign and malignant tumours.


Weight and size were originally the most accepted diagnostic criteria of malignancy in ACC7; their relevance has been confirmed in recent papers.8 9 Using appropriate cut-offs of weight (50 g) and size (6.5 cm), these criteria have a sensitivity and specificity of 91% and 100%, respectively, in an adult patient population.9 However, the role of size in predicting malignancy has been called into question, given the identification of smaller adrenal nodules by new imaging techniques.1012

The differential diagnosis of carcinomas from adenomas has been based on several microscopic features, none of which is an absolute criterion of malignancy. Different scoring systems for ACC have therefore been developed, using mathematical models or numerical scores based on the association of a given threshold for each considered parameter to malignancy.1316 Unfortunately, scoring systems are sometimes difficult to apply and/or time consuming; it is not therefore surprising that several attempts to simplify the diagnostic algorithms for ACC and provide clinically relevant information have been made over the years.9 12 17 18 Current diagnostic parameters include architectural and cytological data together with information on the invasive properties of the tumour. Nine of these have been combined in the Weiss system, which includes three parameters related to tumour structure (“dark” cytoplasm, diffuse architecture, necrosis), three to cytological features (atypia, mitotic count, atypical mitotic figures), and three to invasion (of sinusoids, veins and tumour capsule).15 16 In addition, other parameters have been considered relevant by other authors, including regressive changes (haemorrhage, fibrosis and calcification), abnormal nucleoli, and broad fibrous bands.13 14

Analysis of pathological parameters

According to the literature data and our experience, some criteria are objective and reproducible (nuclear grade, mitotic count, atypical mitotic figures and non-clear cytoplasm), whereas others (diffuse architecture, necrosis, sinusoidal or venous or capsular invasion, regressive changes, abnormal nucleoli, pleomorphism and broad fibrous bands) are more difficult to apply due to their subjectivity (both qualitative and quantitative) and/or their dependence on tumour sampling adequacy.

Based on the original Weiss study,15 high nuclear grade is identified when grades III and IV of Fuhrman are recognised, based on oval or pleomorphic nuclear shape, size larger than 20 μm, coarsely granular chromatin and prominent (macro)nucleoli.19 Mitoses are generally screened counting 10 random fields in each of five slides. An atypical mitotic figure is any cell division having an abnormal polarisation of the condensed chromatin during the different phases of the mitotic process, or more than two mitotic spindles. Eosinophilic cytoplasm is easily described as finely granular, oxyphilic non-clear cytoplasm which must be identified in at least 75% of tumour cells.

Evaluation of the other criteria is, in our experience, less reproducible and would require the use of strict rules. For example, the presence of necrosis, pleomorphism, abnormal nucleoli or broad fibrous bands is not per se hard to recognise, but the identification of these features in small cell groups or in a limited tumour area may equivocally be scored as either present or absent. In particular, it has been suggested that necrosis is regarded as “present” when observed in confluent nests of cells, while for the remaining parameters above there is no internationally accepted definition for their assessment.5 6 The recognition of regressive changes (haemorrhage, fibrosis and calcification) was considered of high value for the diagnosis of ACC in the van Slooten system (being assessed a numerical score of 5.7 with a threshold of ⩾8 for a diagnosis of malignancy) despite the difficulty of their objective evaluation.14

A particularly hard task is the assessment of diffuse architecture. This has been defined as patternless sheets of cells interrupted by a fine sinusoidal network in at least one third of tumour.6 15 A reproducible assessment of diffuse architecture requires a knowledge of what is meant by non-diffuse. Several reports have indicated that non-diffuse patterns included cell arrangements in trabeculae, cords, alveoli or nests. Because of the low threshold for a diffuse architecture (33% of tumour), and the possibility that a compact growth of cords and nests may simulate a diffuse pattern, this criterion is, in our opinion, challenging to apply and poorly reproducible, despite its inclusion in all three major scoring systems.1315 Additional histochemical stains, such as reticulin staining, might be required to correctly assess its presence.

Finally, capsular penetration has been precisely defined as presence of cords and nests of tumour cells inducing a stromal reaction within the capsular tissues,15 but its recognition may sometimes be hampered by either the extremely limited extent of such penetration or the lack of clear cut stromal reaction. The correct classification of venous versus sinusoidal invasion is even more complicated. In endocrine pathology in particular, vascular embolisation by tumour cells is a widely accepted criterion of malignancy, and, for example, is sufficient to prompt a diagnosis of thyroid follicular carcinoma. In adrenocortical tumours, this feature is not sensitive enough for a diagnosis of malignancy, if used alone, but has a relevance in the context of multi-parameter scoring systems. Weiss decided to consider sinusoid and venous invasion separately; an apparently clear-cut distinction between sinusoids (endothelial lined vessels with little supportive tissue) and veins (vessel wall with a smooth muscle component) was made.15 As correctly stated by McNicol in a recent commentary,20 the separation of sinusoidal from venous invasion is difficult and prone to inter-observer variability. More strictly defined rules for venous and sinusoidal invasion were proposed by Aubert et al; these will probably be helpful to provide the appropriate evaluation of each single case.9

Paediatric tumours

Paediatric adrenocortical tumours are even rarer than their adult counterparts and usually display a less aggressive clinical behaviour. Very few studies have addressed this issue in sufficiently powered cohorts, with the AFIP series being the largest on record (n = 83).17 Nine macroscopic or microscopic parameters were found to be associated with malignant behaviour. Among the classical criteria considered in the Weiss system, only confluent necrosis, capsular or vascular invasion, presence of atypical mitotic figures and mitotic count >15/20 high power fields were significantly related to poor prognosis. Four other parameters had a relevant impact on prognosis: tumour weight >400 g, tumour size >10.5 cm, vena cava invasion and periadrenal tissue infiltration.17 From this study it can be concluded that some features important for adult ACC diagnosis (eosinophilic cytoplasm, diffuse architecture or sinusoidal invasion) were apparently less relevant in the paediatric population, indicating that the diagnostic work-up of children’s tumours should probably employ a different panel of pathological factors.

Oncocytic tumours

This term refers to rare adrenal tumours with a predominant component of large, deeply eosinophilic, mitochondrion-rich cells. Forty-five oncocytic ACCs are on record in the literature.2123 The diagnosis of malignancy in oncocytic tumours is difficult using the traditional scoring systems.17 24 However, at least some parameters (eosinophilic cytoplasm, high nuclear grade and diffuse architecture) are intrinsically present in this tumour type.2325 Therefore, the cut-off values validated for conventional ACC may lead to an over-diagnosis of malignancy. It has been suggested that only some of the parameters included in the Weiss system (mitotic activity, necrosis, vascular and capsular invasion) should be taken into account. However, the cut-offs for a diagnosis of malignancy remain unclear.25 Other authors indicated large size, capsular or vascular invasion and surgical un-resectability as features of malignancy, but did not specify the number of features to be identified.26 In a recent study on oncocytic adrenocortical tumours, malignancy was diagnosed in the presence of a “major criterion” (>5 mitoses/50 high power fields, atypical mitosis, venous invasion). Conversely, if a tumour displays only “minor criteria” of malignancy (size >10 cm, weight >200 g, necrosis, capsular or sinusoidal invasion), they are best diagnosed as “of uncertain malignant potential” or “borderline tumours”.23

Myxoid changes

Among other rare morphological features encountered in ACC, myxoid changes are noteworthy, due to the problems they pose for the differential diagnosis with other extra-adrenal myxoid lesions. Focal myxoid extracellular material, highlighted by positive Alcian blue staining, is not uncommon in adrenocortical benign and malignant neoplasms, although on very rare occasions it might represent the majority of the lesion, even in the absence of areas of typical ACC.27 In this context, especially in biopsy material, extra-skeletal myxoid chondrosarcoma, chordoma, myxoid carcinomas from the kidney or other sites, myxoid lipomatous or nerve sheet tumours should be entertained as potential differential diagnoses.

A diagnostic algorithm

Despite revisions and updates,8 9 16 the diagnostic work-up of adrenocortical tumours is still time consuming and poorly reproducible. A better understanding of the “weight” of individual parameters is probably required. In this respect, it is interesting to note that the vast majority of published studies on ACC diagnostic criteria found mitotic count to be the most relevant parameter for a diagnosis of malignancy (table 1).

Table 1 Summary of clinicopathological parameters analysed and found of relevance in the diagnosis of adrenocortical carcinoma

Nevertheless, it was not until 2007 that a study suggested selecting malignant cases through a diagnostic algorithm centred on mitotic count (>5/50 high power fields), followed by the recognition of other parameters (including nuclear grade, diffuse growth pattern and tissue reaction).28 A comparison with the other known scoring systems indicated that a stepwise procedure was the most specific, while the pathological scores proposed in the Weiss and van Slooten systems were the most sensitive for ACC diagnosis. Although this proposal was derived from statistical analysis of relatively few malignant cases, the advantage of this method is to simplify the use of all known diagnostic features, therefore reducing the inter-observer variability in the evaluation of individual criteria.

Morphological versus clinical malignancy

A major issue is the distinction between morphologically and clinically malignant adrenocortical tumours. The diagnostic significance of current pathological parameters has been confirmed by the malignant behaviour of ACCs. Interestingly, some of them have been shown to have prognostic significance.29 30 In a recent report by Assie et al on 124 metastatic cases,31 the importance of mitotic index and the presence of atypical mitotic figures as factors significantly correlated with malignant behaviour and survival was stressed. As described above, in paediatric adrenocortical neoplasms, necrosis, capsular or vascular invasion, presence of atypical mitotic figures and mitotic count >15/20 high power fields were significantly related to poor prognosis, together with tumour weight >400 g, tumour size >10.5 cm, vena cava invasion and peri-adrenal tissue infiltration.17

Immunophenotype and differential diagnosis between ACC and other adrenal and extra-adrenal lesions

Immunohistochemistry is generally not relevant in routine diagnosis to distinguish benign from malignant adrenocortical tumours, despite some proposed markers that are discussed later. However, the differential diagnosis between ACC and other adrenal and extra-adrenal tumours showing similar morphological features, including phaeochromocytomas/paragangliomas, renal cell and hepatocellular carcinomas and metastatic tumours, should be considered. Alpha-inhibin and melan A antibodies are the most sensitive markers for determination of the adrenocortical origin of ACC, although they are not highly specific. Compared to other types of carcinoma, ACC is usually negative or only focally positive for different types of cytokeratins and EMA. Negative staining for CD10 and anti-hepatocyte antigen militates against a renal and liver neoplasm, respectively. Positivity for neuroendocrine markers is restricted to synaptophysin, whereas chromogranin A is invariably negative, a feature that helps in the differential diagnosis with phaeochromocytoma.


In recent years, molecular characterisation of adrenocortical tumours has provided a wealth of data, some of which has already been translated to the diagnostic pathology armamentarium. In order to identify reliable diagnostic and prognostic factors, a wide array of genetic, molecular and immunohistochemical markers have been tested, the most relevant of which are summarised below and in fig 1.

Figure 1 Schematic diagram of the most relevant molecular and phenotypic markers described in adrenocortical carcinoma. Light grey shaded areas highlight possible correspondence at genetic and protein/mRNA expression levels, on chromosomes 11 and 17.

Chromosomal aberrations and allelic imbalance

Several studies have shown that genomic alterations are more prevalent in malignant than in benign and hyperplastic adrenocortical lesions. Numerical chromosomal aberrations (gains, losses and amplifications) can be detected with chromosomal or microarray based comparative genomic hybridisation (CGH) and loss of heterozygosity/allelic imbalances with allelotyping techniques.6 The most common gains in ACC are reported to affect chromosomes 4, 5, 12 and 19, while losses are often found on chromosomes 1, 2, 3, 4, 6, 9, 11, 13, 15, 17, 18, 22 and X.6 32 An aneuploid DNA pattern has often been associated with such chromosomal imbalances, although the value of DNA ploidy analysis is limited for both diagnostic and prognostic purposes.33 34 Kjellman et al35 found a strong relationship between the number of genetic aberrations detected using CGH and tumour size. No alterations were seen in adenomas smaller than 5 cm, whereas an increased number of genetic alterations in the two largest adenomas (5 cm each) and seven of the eight cancers (7–20 cm) were detected.

Extensive genomic imbalances were encountered in ACC by comparative genomic hybridisation, indicating that the molecular pathogenesis of sporadic tumours is complex and that multiple genetic changes drive malignant transformation and tumour progression. A role of tumour suppressor gene(s) (TP53, HIC1, ABL) inactivation on chromosome 17p has been recognised as an early event.32 36

Alterations of genes involved in inherited cancer syndromes

The majority of ACC are sporadic; however a link between ACC and familial cancer syndromes, including the autosomal dominant Li–Fraumeni and Beckwith–Wiedemann syndromes, has been reported.

In Li–Fraumeni syndrome, patients may develop ACC and also a number of other neoplasias, including sarcomas, breast cancer, brain tumours, leukaemia and lymphoma, as a result of germline mutations of TP53 gene and subsequent loss of heterozygosity (LOH) at 17p13.1 locus.3739 Common mutations include substitution of Arg to His at codon 175 (which codes for amino acids of the protein binding site to DNA), and Arg to His at codon 337 (R337H) (coding for the protein oligomerisation domain) observed in children of southern Brazil, where the incidence of ACC is 10–15 times higher than in the rest of the world.40 41 Somatic mutations of TP53 gene, as seen in Li–Fraumeni syndrome, as well as p53 protein accumulation can be detected in sporadic ACC and have been considered as marker of malignancy.4244

In Beckwith–Wiedemann syndrome, various developmental abnormalities can be associated with paediatric tumours, including ACC, as the result of 11p15.5 chromosomal region alterations; several relevant genes may be involved, including IGF2, CDKN1C and H19.45 IGF2 gene is normally expressed from the paternal allele only, while the maternal one is imprinted by promoter methylation. The loss of maternal imprinting or paternal isodisomy induce over-expression of IGF-2, a hormone predominantly expressed during fetal life, which binds to IGFR1 and leads to cell proliferation, differentiation and survival.46 47 IGF2 gene alterations have been described in both familial Beckwith–Wiedemann syndrome and sporadic ACC cases, with a 100-fold IGF-2 over-expression in 60–90% of carcinomas.48 49 Concurrent changes in other genes at 11p15 locus (TP57 and H19) are likely key contributors to ACC development, as suggested by the high prognostic value of 11p15 LOH in ACC.50

Other inherited cancer syndromes associated with adrenocortical tumours (more often adenomas than carcinomas) are multiple endocrine neoplasia type 1 (MEN1), whose locus (11q13) is commonly lost in sporadic adrenal tumours (up to 100% in ACC), although MEN1 gene mutations are extremely rare,5153 and Gardner’s syndrome caused by inactivating mutations of APC gene (on chromosome 5q21), which may be responsible for altered WNT pathways observed in ACC.54

Other genetic and epigenetic abnormalities

LOH at 9p21 associated with lack of p16 protein (a cyclin kinase inhibitor) expression has been found in a fraction of ACCs but not adenomas.55 The role of the zinc finger transcription factors GATA-4 and GATA-6 has been extensively investigated in human cancer tumourigenesis. These factors bind to consensus DNA sequences (A/T)GATA(A/G) in gene promoters and enhancers, and are usually over-expressed (GATA-4) or reduced (GATA-6) in tumour cells. A functional link between luteinising hormone receptor (LHR) and GATA-4 effects in the adrenal pathophysiology has been proposed, and altered up-regulation of LHR and GATA-4 and down-regulation of GATA-6 have been described in ACC.56 57

Gene expression profile

In the few reported studies which applied gene microarray technology, different gene expression profiles between adenomas and ACCs have consistently been found. IGF2 gene up-regulation is one of the most specific molecular features of ACC,49 58 59 a feature which correlates with the altered imprinting of IGF2 locus as outlined above. Among other genes differentially expressed between adrenocortical adenomas and carcinomas, osteopontin (SPP) and serine threonine kinase (STK) 15 were significantly up-regulated in the latter group.49 A gene expression profile similar to adult ACC cases was found in paediatric adrenocortical tumours.60 Moreover, de Fraipont et al showed that 14 genes are strongly associated with tumour recurrence and aggressive behaviour.58

Protein markers

The proliferative activity, as detected by Ki67 labelling index, is considered a relevant marker of malignancy in adrenocortical tumours, with a cut-off value between adenomas and ACCs varying from 1.5% to 4%, according to different authors.9 11 12 24 6163 Among other immunohistochemical markers, the cyclin kinase inhibitor p21 as well as the epidermal growth factor receptor have been found to be over-expressed in ACC.12 6466 Unfortunately, appropriate cut-off expression levels in benign and malignant conditions have not been reported for these markers, thus limiting their use in the routine diagnostic work-up of adrenocortical tumours.

Other less common markers tested in ACC include insulin growth factors (IGF-1 and IGF-2), vascular endothelial growth factor (VEGF), gelatinase, matrix metalloproteinase (MMP) and others. IGF-1 and IGF-2 have been found by immunohistochemistry to be over-expressed in malignant tumours.63 67 By in situ hybridisation, gelatinase A (MMP-2) mRNA was found to be over-expressed in stromal cells (but not in tumour cells) of ACC, but not adenomas.68 Conversely, MMP-2 serum levels were not significantly different in the two tumour types (as opposed to high VEGF and MMP-3 levels in patients with ACC).69 70 A peculiar MMP-2 protein expression in ACC cells (but not in adenomas) has been recently reported as a useful adjunct for ACC diagnosis and a predictive factor of unfavourable clinical evolution.71

Finally, up-regulation of topoisomerase II alpha, down-regulation of N-cadherin and increased telomerase activity have been preferentially reported in malignant adrenocortical tumours.43 7274


Molecular analysis has had a significant impact on the understanding of pathogenetic mechanisms of ACC development and the evaluation of prognostic and predictive markers. Unfortunately, no marker so far described has proven to be sufficiently sensitive and specific to be added to or to replace current morphological parameters used to diagnose ACC. The molecular classification of ACC is certainly an issue of paramount importance, in particular if individualised treatment for patients with ACC is to be achieved.75 Notwithstanding the developments in the field, ACCs remain diagnosed based on morphological criteria and these should still be used for any further analysis and treatment strategy. In fact, the morphological criteria briefly outlined above still have a major role in the diagnosis of ACC. From a practical point of view, the Weiss system is still the easiest to use and the most popular. Nevertheless routine use of at least some of these criteria is not trivial and the need of a more objective system should be the way forward. Such an approach is of very limited applicability in pre-surgical samples, in spite of the increasing use of needle biopsies in the diagnosis of adrenal masses.

We anticipate that new studies to further simplify the diagnostic algorithm for ACC, focusing on reliable individual parameters (mitotic count) should be performed. The ultimate goal of this exercise should be less cumbersome, more objective and accurate methods for the diagnosis of ACC. This would undoubtedly be of great benefit for the design of studies to unravel not only the pathogenesis and the molecular features of adrenocortical tumours, but also to identify novel therapeutic approaches for patients with this aggressive disease.

Take-home messages

  • Adrenocortical carcinoma (ACC) is a rare and aggressive tumour, with a mean survival of less than 3 years.

  • Its pathological diagnosis is difficult, due to the lack of clear-cut morphological criteria for malignancy; scoring systems have been proposed, although some of the proposed criteria are not straightforward and there is a need for standardisation and simplification.

  • Data generated from molecular characterisation have an impact in understanding pathogenetic mechanisms of ACC development and also in evaluating prognostic parameters of relevance for therapeutic purposes. Unfortunately, no marker so far described has proved sensitive and specific enough to be added to or to replace the morphological parameters currently used to diagnose ACC.



  • Competing interests: None.

  • Funding: This work was partially supported by grants from the Italian Ministry of University and Research (MIUR, Rome, ex 60% to MP and MV) and from the Regione Piemonte (Progetto Ricerca Sanitaria Finalizzata, D.G.R. n. 35-4231, 06.11.2006 to MV).