The genomics of undifferentiated sarcoma of soft tissue: Progress, challenges and opportunities

Abstract

Undifferentiated sarcoma of soft tissue (USTS) are aggressive sarcomas that remain a diagnosis of exclusion and show extreme genomic complexity. Many advances in diagnostic criteria have resulted in several revisions in the definition of this rare cancer type. Recent sequencing efforts have illuminated the nature of the genome complexity and have revealed extensive copy number heterogeneity and multiple evolutionary patterns of development. This review places these recent advances into their historical and translational context and covers the changes in nomenclature, molecular classification, and the promise of personalised medicine.

Introduction

Sarcomas are rare cancers of mesenchymal origin and can occur anywhere in the body but are predominantly located in soft tissues and bone. Soft tissue sarcomas have an age adjusted incidence rate of 45 per million per year and represent less than 1 % of all malignant tumours [1]. They comprise roughly seventy subtypes with many displaying distinctive histologies, molecular genetic underpinnings, divergent natural histories and clinical outcomes [2]. The historical pathological classification of sarcomas is based on histological observations based on the shapes of cells (round, spindled, pleomorphic and epithelioid), patterns of growth (storiform, fascicular, plexiform, patternless) and the nature of the extracellular matrix (myxoid, fibrous,fatty). When this information was considered in conjunction with the patient’s age and sex, the location of the tumour and clinical history a rudimentary classification was born. The concepts of differentiation and dedifferentiation were established by cell biologists in the 1800s and pathologists in the main were able to distinguish specific cell differentiation phenotypes by light microscopy alone e.g. smooth or skeletal muscle, nerve tissues or melanocytes but it was only when the ability to probe the differentiation patterns in cells became mainstream through the use of immunohistochemistry that pathologists began to use this information to improve upon the classification. With these tools, pathologists had the ability to, for example, distinguish amongst the spindle cell sarcomas which were of smooth muscle, skeletal muscle or neural differentiation where cell phenotypes were atypical. Importantly they were now able to identify other cancers masquerading as sarcomas such as sarcomatoid carcinomas or melanomas. Moreover, as cytogenetics and other molecular biological techniques were being used to understand the genetic underpinnings of cancer, it was found that many sarcoma subtypes harboured gene translocations which were prototypical of particular entities and could thus be used as diagnostic tools [[3], [4], [5]].

Groups of sarcomas which did not demonstrate a particular line of differentiation and that were difficult to classify were all included under the rubric of malignant fibrous histiocytomas (MFH) which was regarded as a distinct sarcoma subtype by many pathologists (discussion below). MFH is no longer regarded as a distinct tumour entity and is obsolete as terminology, the more current diagnostic appellation being undifferentiated sarcoma of soft tissue (abbreviated here as USTS) which is an umbrella term that incorporates multiple undifferentiated morphological subtypes [2] including undifferentiated pleomorphic sarcoma (UPS). USTS is now considered a diagnosis of exclusion and accounts for 10 % of adult soft tissue sarcomas and most commonly occurs in patients over the age of 50 but the age range is quite wide [2]. The lower extremities are the most commonly affected sites and USTS are frequently deep-seated lesions within muscle but can be located above the fascia (Fig. 1A). USTS are a group of high-grade tumours frequently demonstrating necrosis and can have a pleomorphic morphology (Fig. 1B and C) and with a local recurrence rate between 19–31 % and a metastatic rate of 30–35 % [6]. The prognosis for these patients is poor, with a median survival for those with advanced, metastatic disease of approximately 12 months [7]. Treatment of localized disease in the majority of cases in the United Kingdom comprises surgical excision with neo-adjuvant radiotherapy. The use of chemotherapy remains a matter of discussion but recent improvements in calculating risk of metastases strongly support perioperative chemotherapy in patients with high risk [[8], [9], [10]].

Until recently, the genetics of USTS has been poorly defined, limited to small series of cases and low resolution techniques such as karyotyping and arrayCGH [11,12]. In the current era of massive parallel sequencing and genome wide multi-omic profiling a clearer picture of the extent of genomic complexity (Fig. 1D), heterogeneity and the variable evolutionary histories of USTS is beginning to be appreciated [13,14]. Whole exome (TCGA) and whole genome sequencing (Steele et al.) efforts have demonstrated differences in mutational frequencies, identification of new cancer driver gene mutations in USTS as well as new ways to classify genomic instability in these tumours [13,15]. It is hoped that this rapid expansion of biological information will have a clear impact in the way that future clinical trials and translational research are performed but there are still many bottlenecks that need to be addressed foremost amongst them are consistent criteria for the diagnosis of USTS and the relative dearth of appropriate preclinical models. This review will summarise the evolving nature of the diagnostic criteria for USTS, and provide an updated view of the compendium of molecular advances in this cancer and how this may relate to cancer evolution. A discussion on how future translational studies could be improved by incorporating comprehensive genomic profiling to improve patient stratification and enable personalised medicine for USTS will also be addressed.

The terminology used for undifferentiated sarcomas has a long and interesting history (Fig. 2). It was first known as MFH which was first introduced in the 1960s by Arthur Purdy Stout who thought it to be histiocytic in origin as it demonstrated histiocyte-like properties such as fibroblast transformation and phagocytic abilities in cell culture [16]. In 1971, Kempson described that MFH comprised multiple variants, some with benign features and others with more aggressive behaviour typified by high local recurrence rates and metastatic potential. One of these historical variants was known as fibroxanthosarcoma, an aggressive tumour with cells that contained foamy cytoplasm and multinucleated giant cells [17].

In early 1980s, MFH became the most commonly diagnosed adult soft tissue sarcoma and constituted multiple different subtypes with most diagnoses being made on morphological grounds on haematoxyln and eosin (H&E) stained slides. The MFH tumours showed a wide variety of histological appearances and were categorised into 5 types: storiform-pleomorphic, giant cell, myxoid (myxofibrosarcoma), inflammatory and angiomatoid. The prototypical pattern was the storiform-pleomorphic type showing admixed areas of pleomorphic cells with cartwheeled storiform areas containing a prominent vasculature. Characteristically, the pleomorphic areas contained numerous large, atypical cells including giant forms with hyperchromatic nuclei and abundant cytoplasm. During this period a revolution in diagnostic histopathology was taking place with the routine use of immunohistochemistry and electron microscopy becoming a staple part of the pathologists’ armamentarium. It soon became apparent to some pathologists that MFH did not express immunophenotypical nor ultrastructural features reminiscent of histiocytic cell neoplasms. In an instructive study, the validity of MFH as a diagnostic entity was interrogated by Fletcher et al. using both immunohistochemistry and electron microscopy on a retrospective series of 159 cases [18]. Only twenty-one cases (13 %) could be classified as MFH. The most common diagnoses were pleomorphic and dedifferentiated liposarcoma, which were discerned thorough identification of lipoblasts and areas of well-differentiated liposarcoma respectively. The other common diagnoses were leiomyosarcoma and non-sarcomatous neoplasms (carcinoma, lymphoma and melanoma). These were distinguished by expert pathology review and using a panel of antibodies that included desmin, smooth muscle actin, HHF-35, myoglobin, S100, CD45 and pancytokeratin [18]. Over time it also became clear that the angiomatoid variant was a distinctive tumour characterised by translocations involving ATF1- EWSR1 or CREB1-EWSR1) [19,20].

There was therefore an increasing awareness that MFH may not be a distinct entity particularly in the late 1990s early 2000s. Many of these tumours were rather subtypes of other sarcomas or even other cancer types, and pleomorphic sarcoma or undifferentiated sarcoma were deemed more appropriate designations for those that were not [21,22]. However, probably reflecting the entrenched nature of the term MFH in the medical community it took at least two decades from the discovery that most MFHs were other tumour types before it was finally removed from the World Health Organisation (WHO) tumour classification [23,24]. In the 2013 WHO classification, the term MFH was removed and UPS was introduced as a soft tissue sarcoma with no identifiable cell line or differentiation [23].

In the 2006 WHO classification [24], this group of tumours was labelled as UPS, however the 2013 WHO classification considered the variable morphological traits that USTS exhibit i.e. spindle, epithelioid, pleomorphic (UPS), round cell and mixed and created an overarching grouping of which these subtypes form a part [23]. The undifferentiated sarcomas of round cell phenotype are morphologically and genetically a separate grouping more closely related to the small round blue cell tumours such as Ewing Sarcoma which are more prevalent in the paediatric setting and are not discussed further in this review.

The distinction of USTS subtypes from other tumours is largely dependent on the clinical scenario, availability of resources and the desire to pursue a specific line of differentiation by the attending pathologist. In routine practice a line has to be drawn between costs, time and the clinical benefits of assiduously distinguishing other sarcoma subtypes or even other types of cancer from USTS. In most instances many high-grade soft tissue sarcomas of adults are subjected to similar treatment regimens so making the distinction between histological subtypes is predominantly for prognostic information or academic interest. However, in the era of precision medicine and genomic medicine, the sarcoma treatment landscape is changing and increasingly patients are being recruited to biomarker-based clinical trials using small molecular inhibitors, antibodies to elicit an immune response, cellular-based studies as well as modified radiotherapy protocols which may require judicious subtyping for recruitment [[25], [26], [27]].

The ability to accurately workup a diagnosis of USTS in pathology depends on the extent of tissue sampling, use of a wide range of antibodies and molecular testing to distinguish these tumours from other more well-characterised sarcomas (Table 1). A common clinical scenario for instance is the distinction that needs to be made between a dedifferentiated liposarcoma (DDLPS) which occurs when a non-lipogenic appearing sarcoma arises on the background of a well differentiated liposarcoma. Here it is important to note that dedifferentiation is the process of a tumour losing its differentiation shared with the tissue it is presumed to arise from, which can be viewed as becoming more immature. Conversely, a tumour classed as undifferentiated lacks any evidence of a differentiated phenotype; morphologically an undifferentiated and dedifferentiated tumour can be indistinguishable. Hence, the non-lipogenic component often mimics USTS but is distinguished from USTS by the identified of MDM2/CDK4 gene amplification [28]. Recent clinical trial data suggest the CDK4 inhibitors may be beneficial for patients with advanced dedifferentiated liposarcomas and the prognosis for patients with DDLPS is superior than for USTS [[28], [29], [30]]. The clinical scenario, particularly many less distinctive (“soft criteria”) features e.g. location of tumour, previous history of cancer, sun exposed skin or age of patient may influence the decision to label a tumour as USTS. Moreover, the recent demonstration by the Cancer Genome Atlas (TCGA) that high grade myxofibrosarcoma shares many common transcriptomic features with UPS may blur the lines even further and more research is clearly required in this regard [15]. It is also noteworthy that up to 25 % of soft tissue sarcomas that are radiation induced fall into the category of USTS [2].