As noted above, atypical lipomatous tumor (ALT) and well-differentiated liposarcoma (WDL) are synonyms , used for identical tumors in different anatomical locations. Accounting for 30% to 40% of all liposarcomas, ALT/WDL is the most common form of liposarcoma encountered in late adult life, reaching a peak incidence during the sixth and seventh decades. Men and women are equally affected, although at certain sites (e.g., groin) there appears to be a predilection for men. In the collective experience of the Armed Forces Institute of Pathology (AFIP) and the Mayo Clinic, 75% of cases developed in the deep muscles of the extremities and 20% in the retroperitoneum, with the remainder divided between the groin, spermatic cord, and miscellaneous sites. ^, Rarely do these tumors develop in the subcutis or miscellaneous parenchymal sites.

Symptoms related to these tumors depend on the anatomic site. Those in the extremities develop as slowly growing masses that are present months or even several years before the patient seeks medical attention, whereas those in the retroperitoneum are associated with the usual symptoms of an intraabdominal mass. Because ALT/WDLs contain a significant component of mature fat, they present on computed tomography (CT) as fat-density masses, with mottled or streaky zones of higher density corresponding to the fibrous or sclerotic zones. They also tend to have less well-defined borders than lipomas ( Fig. 14.10 ). ALT/WDL of the esophagus typically present as large, polypoid masses which may even be regurgitated by the patient; such lesions have historically been labeled “giant fibrovascular polyps” of the esophagus but are now known to essentially always represent ALT/WDL instead. ^,

CT scan of atypical lipomatous tumor/well-differentiated liposarcoma (ALT/WDL) of the abdominal cavity and retroperitoneum. The mass, with a low attenuation value, replaces abdominal contents.

Grossly, ALT/WDLs are large, multilobular lesions that range in color from deep yellow to ivory ( Fig. 14.11 ). Many could be mistaken for a lipoma, except for their extremely large size and their tendency to have fibrous bands, gelatinous zones, or punctate hemorrhage. The well-developed lobular growth pattern of benign lipomas is often absent as well.

A, Some ALT/WDL closely resembling normal fat, except for the presence of fibrous bands. B, Others have a more gelatinous appearance.

ALT/WDLs have traditionally been divided into three subtypes: (1) lipoma-like, (2) sclerosing, and (3) inflammatory. Because many ALT/WDLs combine features of both lipoma-like and sclerosing subtypes, the distinction between these two types is often arbitrary, and of limited clinical importance, although it has been suggested that the sclerosing subtype might have a somewhat worse prognosis. These lesions are rarely subclassified in daily practice, although the terms serve to draw attention to the range of appearances that these tumors may assume. In the typical lipoma-like ALT/WDL, the tumor consists predominantly of mature fat with a variable number of spindled cells with hyperchromatic nuclei and multivacuolated lipoblasts ( Figs. 14.12 to 14.15 ). In some cases, these atypical, spindled cells are numerous, whereas in others, the cells are so rare as to require extensive sampling of the tissue. Lipoma-like forms of ALT/WDL will often contain abnormally configured, thick-walled blood vessels containing similar hyperchromatic cells within their walls, a useful clue to this diagnosis in cases with only small numbers of hyperchromatic stromal cells. Sclerosing forms of ALT/WDL, most common in the groin and retroperitoneum, have dense fibrotic zones alternating with mature adipocytes ( Figs. 14.16 to 14.19 ). In some cases, the fibrotic zones consist of trabeculae intersecting fat, and in others the fibrous areas consist of broad sheets. The fibrotic areas contain collagen fibrils of varying thickness embedded with scattered spindled and multipolar stromal cells with hyperchromatic nuclei. Similar cells may also be present between the mature adipocytes. Although lipoblasts may be present, they are usually rare or even absent. Therefore, the diagnosis for this sclerosing pattern of liposarcoma depends more on the identification of stromal cells with a requisite degree of atypia than on the identification of diagnostic lipoblasts. The inflammatory form of ALT/WDL occurs almost exclusively in the paratestis and retroperitoneum and consists of a dense lymphocytic or plasmacytic infiltrate superimposed on a lipoma-like or sclerosing form of ALT/WDL ( Fig. 14.20 ). Because of the intense inflammatory infiltrate, these tumors may be confused with lipogranulomatous inflammation or with hematopoietic neoplasms. Indeed, we have seen cases that were diagnosed repeatedly as “atypical lymphoproliferative processes,” only to be recognized as inflammatory WDL after dedifferentiation. Thus, the diagnosis of inflammatory WDL should be considered for unusual fibroinflammatory processes, especially in the retroperitoneum and paratestis.

Lipoma-like ALT/WDL showing only a rare atypical stromal cell amid a mature lipomatous backdrop.

( A ) Atypical stromal cell in an ALT/WDL illustrating nuclear hyperchromatism. ( B ) Nuclear MDM2 immunostaining in atypical cells in ALT/WDL.

ALT/WDL with a larger number of atypical stromal cells and lipoblasts than in Fig. 14.13 .

ALT/WDL with numerous lipoblasts.

Sclerosing ALT/WDL showing some degree of mature fat ( A ) and no fat ( B ). ( C ) Myxoid pattern in ALT/WDL showing wide separation of fusiform/spindled cells in a stroma having coarse but elaborate vasculature. Areas such as this can be mistaken for myxoid liposarcoma.

Sclerosing ALT/WDL showing sheet-like areas of collagen and fat. Note the multivacuolated lipoblasts in this lesion. These cells are typically rare.

Sclerosing ALT/WDL ( A ) with fibrous bands containing atypical cells ( B ).

Lipoblasts in ALT/WDL.

ALT/WDL of the inflammatory type with a dense lymphocytic infiltrate ( A ) and areas of lipoblastic differentiation ( B ).

ALT/WDLs infrequently display areas of relatively mature smooth muscle ( Figs. 14.21 and 14.22 ). These “lipoleiomyosarcomas” are dual-lineage sarcomas in which both the lipomatous and the smooth muscle components are low-grade. Biologically, their behavior is identical to ordinary ALT/WDL, including the ability to dedifferentiate, and are recognized by areas of ALT/WDL blending with fascicles, nodules, or broad expanses of smooth muscle tissue having mild to moderate nuclear atypia and low levels of mitotic activity. In some cases, the smooth muscle appears to extend out from the walls of large vessels, which similarly contain atypical smooth muscle cells ( Fig. 14.22 ). The amount of smooth muscle varies considerably from case to case, with some tumors showing only occasional foci and others broad expanses. In our experience, MDM2 amplification can be demonstrated in both the adipocytic and smooth muscle components, confirming that this smooth muscle component is neoplastic, rather than some unusual reactive phenomenon. A less common association is ALT/WDL with low-grade osteosarcoma-like areas. These tumors, occurring primarily in the retroperitoneum, consist of a lipomatous component, which usually predominates and blends with areas resembling parosteal or low-grade intramedullary osteosarcoma. These areas are characterized by relatively mature bone, with or without osteoblastic rimming, embedded in a low-grade fibroblastic backdrop. In a few cases, high-grade osteosarcoma coexists in the low-grade areas. In the few cases studied, MDM2 and CDK4 have been demonstrated immunohistochemically in both the lipomatous and osteosarcomatous components.

ALT/WDL with smooth muscle differentiation (lipoleiomyosarcoma) ( A ) and stained with Masson trichrome stain ( B ).

ALT/WDL with smooth muscle differentiation (lipoleiomyosarcoma) showing atypia in a vessel wall.

Occasionally, ALT/WDLs have a predominantly myxoid appearance, a phenomenon that has led to the misclassification of some of these tumors as either a variant of myxoid liposarcoma or a mixed type of liposarcoma (see Fig. 14.16C ). In these areas, the cells are more spindled, and the vessels coarser than in a myxoid liposarcoma. Several studies have shown that these tumors lack DDIT3 rearrangements and therefore are unrelated to myxoid liposarcoma. ^, It is extraordinarily rare for a primary myxoid liposarcoma to arise in the retroperitoneum (see below).

ALT/WDL is characterized by giant marker and ring chromosomes, sometimes as a sole finding or occasionally in association with other numeric or structural alterations. The giant marker and ring chromosomes contain amplified sequences of 12q13-15, the site of several genes (e.g., MDM2, GLI 1, TSPAN31, CDK4 , HMGA2 ). MDM2 (12q13-14) and HMGA2 (12q15), part of the same amplicon, are consistently amplified as a result of this abnormality. CDK4 , located at 12q13, and TSPAN31 , located at 12q13-q14, belong to a separate amplicon, which is coamplified with MDM2 and HMGA2 in about 90% of cases. GLI1 and DDIT3 (12q13.1-q13.2) are infrequently amplified. Amplification of MDM2 and CDK4 results in downstream signaling, the net result of which is to inhibit apoptosis and increase cell proliferation. MDM2 binds to p53, thereby decreasing apoptosis, whereas CDK4 phosphorylates the RB1 gene product, preventing its interaction with E2F transcription factor and allowing the cell cycle to escape the G1-S checkpoint.

Immunostaining for CDK4 and MDM2 is a reasonable first-line tool for separating ALT/WDL from various benign lipomatous lesions. ^, Immunoreactivity can be detected within the majority of ALT/WDLs and DLs; the exact percentage varies depending on the study. MDM2 staining is localized to the nuclei, where it is most easily visualized in the large hyperchromatic cells (see Fig. 14.13B ). The percentage of positive cells also varies from case to case. Immunoreactivity is not generally present within deep lipomas. However, a small percentage of spindle cell/pleomorphic lipomas contain occasional cells positive for MDM2 and/or CDK4. ^, In addition, antibodies to CDK4 and MDM2 occasionally stain nuclei of histiocytes within areas of fat necrosis, making it imperative that the character of the immunopositive nuclei be evaluated. Nuclear MDM2 and CDK4 can also be detected by immunohistochemistry (IHC) in a small subset of non-lipomatous sarcomas (e.g., malignant peripheral nerve sheath tumor). IHC for p16 has also been shown to be of value in the distinction of ALT/WDL from lipoma, although this seems to be less sensitive than IHC for MDM2.

In contrast, MDM2 gene amplification evaluated by fluorescence in situ hybridization (FISH) is a highly sensitive and specific means of diagnosing ALT/WDL, ^{, ,} even in needle biopsy material. It is significantly superior to immunostaining, which not only fails to identify all ALT/WDLs, but also is associated with a small but definite false-positive rate. Virtually all ALT/WDLs display amplification of MDM2 in both biopsy and resection specimens. MDM2 is not amplified in lipomas, although spindle cell/pleomorphic lipomas may display polysomy of the 12q locus. Because of its exquisite sensitivity and specificity in the diagnosis of ALT/WDL, FISH has been endorsed for the evaluation of ambiguous (vs. obvious) lipomatous lesions. Most often, these are tumors in which the degree of atypia falls short of a threshold level for the diagnosis of malignancy. Admittedly, this threshold is different from pathologist to pathologist, although, as a group, pathologists tend to overestimate the degree of atypia present when contemplating a diagnosis of ALT/WDL. The situations in which molecular testing for MDM2 is highly recommended are lipomatous tumors with equivocal cytologic atypia, recurrent lipomas, deep lipomas without atypia that exceed 10 cm, and retroperitoneal or intra-abdominal lipomatous tumors lacking cytologic atypia ( Box 14.1 ). ^, These guidelines are especially important in evaluating lesions in patients older than 50 years, since the risk of liposarcoma increases with age. Using this approach, reasonable evidence now exists to recognize the entity of retroperitoneal lipoma, a diagnosis formerly considered to be hearsay. Nevertheless, although its existence has been established, retroperitoneal lipoma is exceedingly rare. The diagnosis should be made only after WDL has been excluded with FISH.

It should also be kept in mind that a very small percentage of well-differentiated and dedifferentiated liposarcomas may harbor amplifications of the MDM4 gene, rather than MDM2 . ^, As would be expected, such tumors are negative for MDM2 amplification by FISH and immunohistochemistry, emphasizing the need to integrate morphological and molecular genetic findings in the differential diagnosis of adipocytic tumors with cytologic atypia. In our experience, MDM4 amplification can be reliably detected by single-nucleotide polymorphism assays, such as the OncoScan assay.

ALT/WDLs are non-metastasizing lesions that are traditionally not graded. However, their rate of local recurrence and disease-related mortality are strongly influenced by location. ^{, , ,} As shown in Table 14.2 , rates of local recurrence for tumors in the extremities (almost 50%) are significantly lower than for tumors in the retroperitoneum (almost 100%). One could argue that ALT/WDL of the retroperitoneum is basically an incurable lesion. About one-third of patients die as a direct result of their disease, but this figure increases with longer follow-up as a result of the indolent growth of these lesions. On the other hand, those rare ALT/WDLs that occur in the subcutaneous tissues are not associated with tumor-related death and are generally cured by limited excision. An interesting series of peripherally located ALT and dedifferentiated liposarcoma, reported by Kalimuthu et al., noted local recurrences in 9% of affected patients, with over 50% of recurrences occurring more than 60 months after diagnosis, and one tumor recurring 40 years after initial surgery. This study emphasizes the need for long-term follow-up of patients with this disease.

ALT/WDLs should not be dismissed as little more than benign but locally aggressive lesions, because a small percentage of these tumors over time will progress histologically to a higher-grade lesion (DL). ^{, ,} Although this phenomenon occurs most frequently with retroperitoneal liposarcomas, it also occurs with deep extremity lesions; it is rare in subcutaneous tumors. Therefore, it does not appear to be a site-specific phenomenon as formerly believed, but rather a time-dependent phenomenon encountered in locations where there is a high likelihood of clinically persistent disease. With retroperitoneal tumors, for which complete excision is virtually impossible, there is a substantial risk of dedifferentiation (about 10% to 15%); it is somewhat lower for extremity lesions (5%). In ALT/WDLs that have been followed longitudinally, dedifferentiation occurs after an average of 7 to 8 years but may be seen as long as 16 to 20 years after the original diagnosis. When dedifferentiation occurs, the lesions can usually be considered fully malignant sarcomas.

Various neoplastic and non-neoplastic lesions enter the differential diagnosis of ALT/WDL ( Box 14.2 ). For most of these conditions, none of the available histochemical or immunohistochemical stains is useful. Rather, careful sampling of the material and thin, well-stained hematoxylin-eosin sections constitute the mainstay of accurate diagnosis. Lipid stains, although obviously positive in ALT/WDL, also disclose lipid-positive deposits in the vast panorama of reactive lesions in fat and a variety of tumors.

• •

  Normal fat with Lochkern cells . Normal white fat consists of spherical cells containing one large lipid vacuole that displaces the thin oval nucleus to one side. On routine sections the nucleus of most fat cells is barely perceptible. From time to time, a section grazes an adipocyte nucleus so that it is viewed en face, displaying its characteristic central vacuole, termed Lochkern (German, “hole in the nucleus”) ( Fig. 14.23 ). Lochkern cells are seen more frequently in thick sections and are sometimes misinterpreted as evidence of lipoblastic differentiation and therefore a liposarcoma.

  

  Intranuclear vacuoles (Lochkern cells) in normal fat.

• •

  Fat necrosis . In areas of fat necrosis, finely granular or vacuolated macrophages are located in the vicinity of damaged fat characterized by diminished cell size, dropout of adipocytes, and chronic inflammation (see Fig. 14.3 ). Unlike lipoblasts, macrophages are of uniform size and have small, evenly dispersed vacuoles that do not indent the nucleus. The nucleus has a rounded shape with delicate staining. In thick sections, the nuclei of macrophages may overlap one another, giving the impression of hyperchromatism , which typifies the atypical stromal cells in ALT/WDL. Lochkern cells are also common in fat necrosis. When making such distinctions, it is important to have suitably thin histologic sections. Macrophages within fat necrosis are routinely MDM2-positive by immunohistochemistry, a potential pitfall.

• •

  Atrophy of fat . Starvation, malnutrition, and local trauma result in atrophy of fat. Atrophy is accompanied by a loss of intracellular lipid such that the cell shrinks dramatically and assumes an epithelioid shape (see Fig. 14.4 ). With loss of lipid, the nuclei become more prominent, and the cells superficially resemble lipoblasts. Important observations include cells that appear to be of uniform size and maintain their arrangement in lobules. With extreme atrophy, the cells may contain lipofuscin. Such changes are particularly noticeable in subcutaneous tissue and the omentum.

• •

  Massive localized lymphedema . Massive forms of lymphedema restricted to a portion of the body may be confused clinically and histologically with ALT/WDL. These lesions develop in morbidly obese individuals and appear to be the result of lymphedema secondary to chronic dependency of a fatty panniculus. Not surprisingly, these lesions develop in the proximal extremities and may be aggravated by underlying factors, such as lymphadenectomy. Grossly and microscopically, the lesions exhibit the changes of lymphedema, including thickening of the overlying skin, dermal fibrosis, ectasia, proliferation of lymphatics with focal cysts, and expansion of connective tissue septa ( Fig. 14.24 ). A misdiagnosis of liposarcoma is attributable to the expanded connective tissue septa that are believed to be part of a sclerosing liposarcoma. The septa contain mild to moderately atypical fibroblasts and delicate collagen fibrils separated by edema. In addition, there is often striking vascular proliferation at the interface between the expanded connective tissue septa and lobules of fat.

  

  Changes of lymphedema that may mimic an ALT/WDL. ( A ) Connective tissue septa are expanded with ( B ) mildly atypical fibroblasts in the septa.

• •

  Silicone reaction . Injection of silicone for various therapeutic and cosmetic purposes results in sheets of massively distended multivacuolated histiocytes that may closely mimic lipoblasts (see Fig. 14.6 ). Lipoblasts of such quality and number are rarely encountered in true liposarcomas. Silicone reactions are also accompanied by a modest inflammatory and giant cell reaction and large cysts with eosinophilic borders. Most silicone reactions in clinical practice are encountered around silicone breast implants but occasionally are seen on the face, abdomen, and genital region. Free silicone can also migrate under gravitational effect and therefore is found at sites distant from the original introduction site.

• •

  Intramuscular lipoma with atrophic skeletal muscle . Infrequently, atrophic skeletal muscle fibers are seen in intramuscular lipomas ( Fig. 14.25 ). When these collections retain a clustered arrangement and have identifiable eosinophilic cytoplasm, this phenomenon is easily recognized. Isolated degenerating myofibers with barely perceptible cytoplasm understandably can be misidentified as atypical stromal cells of ALT/WDL. Positive identification can be accomplished with desmin immunostains.

  

  Atrophic skeletal muscle in an intramuscular lipoma. Degenerating myofibers are occasionally mistaken for atypical cells in liposarcomas.

• •

  Herniated orbital fat . Prolapse of subconjunctival intraconal fat is a rare cause of an intraorbital mass. Herniated orbital fat, unlike normal orbital fat, contains floret-type giant cells, a feature that often leads to a mistaken diagnosis of liposarcoma ( Fig. 14.26 ). This condition develops in adults in the region of the superotemporal quadrant of the orbit or the lateral canthus below the lacrimal gland. The lesion can be unilateral or bilateral. The nuclei of the multinucleated giant cells present in herniated orbital fat are small, round, and normochromatic, unlike those of ALT/WDL.

  

  Herniated orbital fat at low power ( A ) and floret-type giant cells at high power ( B ).

• •

  Sclerosing extramedullary hematopoietic tumor (SEMHT) . This rare pseudoneoplasm, seen in patients with chronic myeloproliferative disorders, such as chronic myeloid leukemia, essentially represents dysplastic extramedullary hematopoiesis, occurring within the soft tissues. When SEMHT involves fat, dysplastic megakaryocytes may closely mimic the enlarged, hyperchromatic stromal cells of ALT/WDL. Clinical correlation with regards to a history of a myeloproliferative disorder, recognition of other hematopoietic elements such as myeloid and erythroid precursors, and occasionally ancillary immunohistochemical studies for markers such as CD61 (megakaryocytes), myeloperoxidase (myeloid cells), and hemoglobin (erythroid precursors) should allow for the correct diagnosis in such patients.

• •

  Perinephric myxoid pseudotumor of fat . This unusual pseudoneoplastic process typically occurs in patients with medical kidney disease, and less often a renal neoplasm, and presents as a large, retroperitoneal mass. ^, Histologically, it consists of an admixture of mature fat, myxoid stroma, a proliferation of generally bland fibroblastic stromal cells, and a mixed chronic inflammatory cell infiltrate, simulating a well-differentiated liposarcoma with myxoid change. Unlike WDL, however, perinephric myxoid pseudotumor of fat lacks enlarged, hyperchromatic stromal cells and is negative for MDM2 amplification. Its pathogenesis is unknown.

Dedifferentiation, or histologic progression to a higher-grade, less well-differentiated neoplasm, was first described as a late complication in the natural history of well-differentiated chondrosarcoma, but it is now known to occur in other low-grade mesenchymal tumors, including parosteal osteosarcoma, chordoma, and ALT/WDL. Dedifferentiated liposarcoma (DL) has traditionally been defined as ALT/WDL juxtaposed to areas of high-grade non-lipogenic sarcoma, usually resembling undifferentiated pleomorphic sarcoma and most often occurring after a latent period of several years. The concept of “low-grade” DL is controversial ^, and will be discussed at length below. In roughly 90% of cases dedifferentiation is present at initial presentation, with the remainder representing recurrences of previously diagnosed WDL.

Dedifferentiated liposarcomas account for roughly 20% of liposarcomas, develop in approximately the same age group as ALT/WDL, and reach a peak incidence during the early seventh decade. ^{, , ,} DL occurring in patients <40 years of age is quite rare, accounting for roughly 3% of cases. Men and women are affected equally. Unlike ALT/WDL, location in the retroperitoneum is favored over deep soft tissues of the extremities by almost 3:1. Fewer than 20% of DLs occur collectively in the head, neck, trunk, and spermatic cord and rarely in the subcutis. Rarely, patients may present with leukemoid blood reactions. Radiographically, DLs have areas characteristic of ALT/WDL but also have mass-like areas of non-fatty tissue. The latter appearance has imaging characteristics similar to other sarcomas, with prolonged T1 and T2 relaxation on magnetic resonance imaging (MRI) and attenuation coefficients higher than those for normal fat on CT scans.

The lesions present as large, multinodular masses ranging in color from yellow to yellow-tan, admixed with firm tan-gray areas that correspond to the dedifferentiated foci. Areas of necrosis, hemorrhage, and cystic change may be present ( Fig. 14.27 ).

Dedifferentiated liposarcoma involving the colon. Areas of well-differentiated liposarcoma ( right ) appear as yellow, homogenous, fatty nodules, whereas the dedifferentiated component ( left ) has a nonspecific, sarcomatous appearance, with a focus of necrosis and cystic change.

In most cases, careful inspection will disclose areas of preexisting ALT/WDL, displaying the range of morphologic features described above, juxtaposed to a non-lipogenic (dedifferentiated) component. The interface between the two zones is typically abrupt ( Fig. 14.28 ), although in some cases there is a gradual transition between the two ( Fig. 14.29 ). Rarely, the two patterns commingle, giving the impression of mosaicism ( Fig. 14.30 ).

Dedifferentiated liposarcoma with sharp abutment of two zones.

Dedifferentiated liposarcoma with indistinct margin between well-differentiated and dedifferentiated zones.

Mosaic pattern of dedifferentiated liposarcoma.

Provided that areas of ALT/WDL are present, the diagnosis of DL is usually quite straightforward. In the great majority of cases, the dedifferentiated zones have the appearance of an undifferentiated pleomorphic sarcoma, consisting of a highly cellular proliferation of clearly malignant, pleomorphic, mitotically active spindled cells growing in sheets, fascicles, and storiform arrays ( Figs. 14.31 and 14.32 ). The dedifferentiated areas can closely resemble high-grade myxofibrosarcoma, with abundant myxoid matrix and a well-developed, arborizing, thick-walled vasculature ( Fig. 14.33 ). Heterologous differentiation may be present, in the forms of osteosarcoma, rhabdomyosarcoma, leiomyosarcoma, or rarely angiosarcoma ( Figs. 14.34 and 14.35 ). Rarely, the dedifferentiated component may consist of highly malignant-appearing mononuclear cells surrounded by a dense neutrophilic infiltrate, identical to the so-called “inflammatory variant” of undifferentiated pleomorphic sarcoma (an MDM2/CDK4 -amplified neoplasm which appears to represent DL in almost all instances). ^, Other unusual findings seen in DL include undifferentiated, large round cells resembling carcinoma or melanoma ^, ( Fig. 14.36 ), “inflammatory myofibroblastic tumor-like” features, and “homologous” lipoblastic differentiation, resembling pleomorphic liposarcoma. ^, The latter tumors were at one time considered to represent “mixed-type” liposarcoma, but demonstrate MDM2 amplification, supporting classification as a DL variant. Finally, small subsets of DL may contain amianthoid fibers ( Fig. 14.37 ), areas closely resembling myxoid liposarcoma ( Fig. 14.38 ), or spindle cells forming whorled structures reminiscent of a meningioma or perineurioma, often in association with spicules of woven bone ( Fig. 14.39 ). It has recently been shown that DL with myxoid liposarcoma-like histology often harbor amplifications of the DDIT3 locus, whereas coamplification of GLI1 is frequently present in those tumors containing meningothelial or perineurioma-like foci.

Dedifferentiated liposarcoma with areas resembling high-grade undifferentiated pleomorphic sarcoma.

Dedifferentiated liposarcoma with areas having the appearance of fibrosarcoma.

Myxofibrosarcoma-like histology in the dedifferentiated component of an MDM2 -amplified retroperitoneal liposarcoma. In this anatomical location, a “myxofibrosarcoma-like” tumor almost always represents dedifferentiated liposarcoma.

Dedifferentiated liposarcoma with rhabdomyosarcomatous differentiation in dedifferentiated areas.

Desmin expression in dedifferentiated liposarcoma with rhabdomyosarcomatous differentiation.

Dedifferentiated liposarcoma composed of undifferentiated, large epithelioid cells.

Dedifferentiated liposarcoma with amianthoid fibers.

MDM2 -amplified dedifferentiated liposarcoma with extensive myxoid change, closely mimicking true myxoid liposarcoma. In contrast to the small, round, and regular cells of true myxoid liposarcoma, the neoplastic cells are enlarged, spindle-shaped, hyperchromatic, and irregular.

Dedifferentiated liposarcoma with areas of whorled structures.

Cellular well-differentiated liposarcoma versus “low-grade dedifferentiated liposarcoma”: A difficult problem without easy answers .

To understand the issues surrounding the concept of “low-grade dedifferentiation,” some historical context is important. Prior to the beginning of the “genetic era,” which for tumors of adipose tissue roughly corresponds to the publication of the seminal CHAMP (Chromosomes and Morphology) study, liposarcomas were considered to represent a single morphologically variable entity, rather than multiple genetically distinct tumor types, as we now know. In 1979, Harry Evans published his seminal study of 55 cases of liposarcoma, rigorously distinguishing them from other sarcoma types (in particular, undifferentiated pleomorphic sarcoma/“malignant fibrous histiocytoma”). Evans subdivided these cases into “myxoid liposarcoma,” “pleomorphic liposarcoma,” “well-differentiated liposarcoma,” and identified 10 cases which he considered to represent variants of WDL, consisting of both WDL and cellular, pleomorphic and mitotically active spindle cell sarcoma, resembling undifferentiated pleomorphic sarcoma, fibrosarcoma, or pleomorphic liposarcoma (an 11th case, considered “unclassified liposarcoma,” seems in retrospect to belong to this group). Drawing an analogy with dedifferentiated chondrosarcoma, Evans termed these “dedifferentiated liposarcomas,” noting that mitotic figures were easy to find in the dedifferentiated zones, equaling or exceeding 5 mitoses/10 high-powered fields in all. This number seems to have been chosen entirely empirically, and there is no mention of similar or “borderline” tumors having a lesser number of mitotic figures. Two of these 11 dedifferentiated liposarcomas metastasized. It is likely that similar cases had been previously reported as “unclassified liposarcomas” or had not been recognized as liposarcomas.

Following Evans’ seminal publication, a number of series of DL were reported, each of which appeared to have used somewhat different, empirically derived criteria for this diagnosis, generally revolving around the presence of anaplastic spindled cells and an abrupt transition from WDL to dedifferentiated areas. ^{, , ,} Weiss and Rao also required that the dedifferentiated focus be macroscopically visible (at least 1 cm). In none of these series was a particular mitotic index required.

In 1997, Henricks and colleagues, in a series of 155 cases, expanded the definition of DL to include cases in which the putative dedifferentiated component was histologically low-grade, “resembling fibromatosis or a well-differentiated fibrosarcoma.” This series empirically defined dedifferentiation as “a region devoid of lipogenic differentiation occupying at least a low-power (x10) microscopic field” and showing “cellularity in excess of that encountered in sclerosing WDL;” no specific level of mitotic activity was required. Of 155 cases, 14 were considered to show only low-grade dedifferentiation; clinical follow-up on 12 of these cases disclosed metastatic disease in 3 patients, 2 of whom died of disease. Importantly, local recurrences (prior to metastasis) in 2 of these patients showed high-grade features, and the third case was incompletely excised, suggesting that high-grade foci might not have been sampled. A subsequent series using identical criteria, by Elgar and Goldblum, identified exclusively low-grade dedifferentiation in 11 of 20 studied cases, only one of whom developed metastatic disease following local recurrences. Unfortunately, the histologic features of these recurrences and metastases could not be studied.

Although the concept of low-grade dedifferentiation in WDL was endorsed in the 2020 World Health Organization classification, specific criteria for this diagnosis are notably absent, with low-grade DL noted simply to be “virtually indistinguishable from cellular well-differentiated liposarcoma.” This definition (or lack thereof) obviously leaves individual pathologists considerable latitude in application, and we have noted over the past several years in our consultation practices marked variation in cases submitted as representing low-grade DL. Adding further to the confusion, the widely used FNCLCC grading system for sarcomas defines DL as representing at least a Grade 2 sarcoma, as this diagnosis is automatically assigned a differentiation score of “3” for purposes of grading (see Chapter 1 ).

Two recent studies have attempted to specifically address these issues. The first of these, also by Evans, examined 61 cases of ALT/WDL and DL with a minimum follow-up duration of 10 years. Four categories of tumors were defined: (1) conventional ALT/WDL, (2) cellular ALT/WDL, having moderately or occasionally highly cellular fibrous or myxoid non-lipogenic regions with <5 mitoses/10 HPF, (3) dedifferentiated liposarcoma, defined as the combination of ALT/WDL and cellular, non-lipogenic sarcoma with at least 5 mitoses/10 HPF, and (4) ALT/WDL with a pleomorphic liposarcomatous component (“homologous” lipoblastic differentiation). Although local recurrences were common in all groups, in particular for tumors located in central body cavity locations, metastases were seen only in the dedifferentiated liposarcoma group. The overall survival of the dedifferentiated liposarcoma group was significantly worse than that of the conventional or cellular WDL cases; the latter categories did not show significant differences in survival.

The second study, by Graham et al., took a similar approach, dividing their series of 98 primary retroperitoneal liposarcomas with median clinical follow-up of 9.3 years into (1) WDL without non-lipogenic areas, (2) WDL with non-lipogenic areas (defined as areas without adipocytes comprising at least a 10× field and at least 50% of one slide) and <5 mitoses/10 HPF, and (3) WDL/DDLPS with non-lipogenic areas and ≥5 mitoses/10 HPF. Local recurrence rates were 23% for conventional WDL, 50% for WDL with non-lipogenic areas and <5 mitoses/10 HPF, and 58% for WDL/DDLPS with non-lipogenic areas and ≥5 mitoses/10 HPF; distant metastases were seen in 0%, 5.6%, and 20%, of these three groups, respectively. By report, metastatic disease occurred only in Group 2 patients whose local recurrences demonstrated non-lipogenic sarcoma with ≥5 mitoses/10 HPF (Dr. Sarah Dry, personal communication). Overall, 27%, 44%, and 70% of patients in these three groups, respectively, died of disease during the study period with median disease-specific survival of 6.7, 8.8, and 5.5 years for these same groups. In general, worse disease-specific survival was seen in subgroups of DL with even higher mitotic activity, but this did not reach statistical significance.

The results of the Evans and Graham et al. studies call into question the existence of low-grade DL or suggest that the prognosis for WDL having non-lipogenic foci with low mitotic activity is favorable, whichever label we choose to apply. There is merit, we think, in not labeling such tumors as representing “DL,” especially in centers where the FNCLCC grading system is utilized, for the reasons discussed above. Conceivably, modification of the FNCLCC system to include differentiation scores of other than “3,” as has been suggested by Riciotti et al., might solve some of these problems, but changes of this type must obviously be made in consensus fashion. We would especially caution pathologists against making the diagnosis of “low-grade” DL in needle biopsies, as it has been our personal experience that the great majority of such tumors ultimately end up being classified as conventional high-grade DL, or less often as cellular WDL, in subsequent larger biopsies or resection specimens.

Despite this progress, certain issues remain. The first of these is how to approach very rare WDL which do truly seem to give rise to histologically bland, morphologically distinct spindle cell tumors, clearly differing from the usual cellular WDL. Such cases are exemplified by instances in which conventional WDL is juxtaposed to large zones of cytologically bland, mitotically inactive spindled cells growing in meningothelial or perineurial-like whorls, often in association with woven bone. ^{, ,} Such tumors clearly deviate from our conventional conception of the morphologic spectrum of WDL and are considered by most (but not all) expert soft tissue pathologists to represent a form of DL. A very recent study by Sharma and coworkers has shed new light on these enigmatic tumors, showing them in most instances to have coamplification of the GLI1 locus (in combination with MDM2 / CDK4 ), a high risk for local recurrence (sometimes with death from disease), but little or no metastatic risk. Should we consider these DL with distinctive genetic findings and especially low metastatic risk, or do they instead represent highly unusual WDL? There are no clear answers, and the need for further study and consensus should be obvious.

There are also issues related to mitotic figure counting. First, although the 5 mitotic figure threshold established by Evans and used in later studies by him and Graham et al. does seem to correspond to behavior and outcome in these tumors, it is entirely empirical, and it is certainly possible that some other cutoff might prove better. Mitotic figure counting itself is highly subjective and subject to sampling variation, and more objective measures such as automated digital analysis for Ki-67 labeling index, possibly in combination with artificial intelligence, might prove more clinically relevant. Finally, as has been noted in prior editions of this textbook, many high-grade sarcomas contain areas with <5 mitoses/10 HPF, and thus mitotic figure counting must be integrated with all available clinical, radiographic, and pathologic information in the distinction of WDL from DL.

Our understanding of the molecular events that determine dedifferentiation is still evolving. Currently, no consistent genetic aberration appears to separate ALT/WDL from DL in a statistically reproducible manner. In fact, the genetic similarity between matched pairs of a given ALT/WDL and its dedifferentiated component indicates that most of the abnormalities present in the lipomatous component of DL are present before phenotypic changes of dedifferentiation occur. On the other hand, pure ALT/WDLs without dedifferentiation have fewer complex abnormalities. These findings could help identify lesions that are at risk to dedifferentiate. At the same time, they do not negate that dedifferentiation is a time-dependent phenomenon occurring after a cumulative series of genetic events.

Generally, DLs display more extensive chromosomal abnormalities than ALT/WDL. The 12q13-15 amplifications are more complex than those in ALT/WDL. Other amplifications, including 1q23, 12q24, and either 6q23 or 1p32, are encountered in about two-thirds of cases. The 6q23 amplicon is the seat of a candidate gene ( MAP3K5 ) that inhibits lipogenic differentiation through c-JUN– or PPAR-γ–dependent pathways. Amplification of STAT6 , with high-level expression of STAT6 protein in tumor cell nuclei, occurs in a subset of DLs and may result in confusion with malignant solitary fibrous tumors. It has recently been suggested that DL showing coamplification of MDM2 and FRS2 may have a somewhat better prognosis than tumors showing a MDM2+/FRS2 − phenotype, although study of a larger number of cases is necessary. Other genes reported to be preferentially amplified in DL as compared to WDL include PTPRB, WIF1, JUN, LRIG3, GLI1, and DDIT3 . Coamplification of DDIT3 and GLI1 have been associated with the presence of myxoid liposarcoma-like histology, meningothelial/perineurial-like whorls, and morphology resembling GLI1 -amplified soft tissue tumors in DL. Subsets of DL also display MDM4 amplification. ^,

There is some evidence to suggest that the behavior of DL may be better than that of other high-grade pleomorphic sarcomas in adults, although this may depend on the studied outcome measure (e.g., metastases vs. survival). This possibility was first suggested by McCormick et al. in a series of 32 DL of various locations, with metastases and death from disease observed in only 13% and 28% of patients with clinical follow-up. Similar findings were seen in the considerably larger series of 155 patients reported by Henricks et al., with local recurrences, metastases, and death from disease reported in 41%, 17%, and 28% of patients, respectively. Evans’ 2007 study of WDL and DL included 24 patients with DL; metastases occurred in 29% and all but one patient died of disease, with a median survival of 77 months. Tirumani and coworkers, in a study of 148 patients with DL in various anatomical locations, found an overall metastatic risk of 30%, with a median survival of 28 months for patients with metastatic disease.

Looking at retroperitoneal DL, Mussi et al. also noted a low metastatic risk (9%), although the 5-year disease-specific survival was only 42%. A similar study of 119 retroperitoneal DL, by Keung et al., found 5-year local recurrence-free survival of only 15%, 5-year distant recurrence-free survival of 33%, and overall 5-year survival of 42%. Gronchi and coworkers, reviewing the Milan experience with retroperitoneal DL, noted distant metastases in only 17.4% of patients, although nearly 60% died of disease. Binh et al., comparing the behavior of retroperitoneal DL and leiomyosarcoma, found DL to have a lower 5-year recurrence-free survival (45% vs. 71%), but notably better 5-year metastasis-free survival (73% vs. 39%). Overall survival was not, however, statistically different between these two groups. Putting all of this together, it seems fair to conclude that the risk for metastases is relatively low for retroperitoneal DL as compared to other pleomorphic sarcomas, although the overall outcome is still quite poor, likely reflecting aggressive and unresectable local recurrences.

Two recent large series have examined the natural history of DL in non-retroperitoneal locations. The first of these, by Tseng et al., studied 58 DL of the trunk and extremities, and found a very low 5-year rate of distant metastases for both FNCLCC Grade 2 (7%) and Grade 3 (12%), and excellent overall 5-year survival (85%). The second, by Morii et al. from the Japanese Musculoskeletal Oncology Group, included 132 patients with DL of the extremities and trunk and found 5-year local recurrence-free, metastasis-free, and disease-specific survival rates of 72%, 76%, and 85%, respectively. These two studies strongly suggest that the natural history of high-grade DL in the extremities and trunk is considerably less aggressive than that of other pleomorphic sarcomas in these locations.

Although there is general agreement that site is the most important prognostic factor for DL, with retroperitoneal tumors having the worst prognosis, there is conflicting data as to other possible prognostic features. Table 14.3 reviews the findings of selected series of high-grade (FNCLCC Grades 2 and 3) DL, with regard to features associated with a higher risk of local recurrence, distant metastasis, and worse overall survival. For the reasons discussed above, prognostic factors related to putative Grade 1 DL are not included, particularly as there is no real evidence that the natural history of “Grade 1 DL” is any different from that of ALT/WDL. Among possible prognostic features, FNCLCC Grade 3 and/or very high mitotic activity (>20/10 HPF) have most consistently been associated with an increased risk of local recurrence, distant metastases, and worse overall outcome. ^{, , ,} Looking only at overall survival, a retrospective study by Goottee and coworkers, of 3,573 patients with DL in the National Cancer Database (USA), found male sex, patient age >50 years, tumor size >10 cm, retroperitoneal/abdominal location, FNCLCC Grade 3, and Stage 3 or 4 disease to be significantly associated with worse overall outcome. Although retrospective database studies of this type have obvious limitations, in particular the absence of central pathology review, these findings seem reasonable.

The most common problem in the differential diagnosis is distinguishing between a fat-infiltrating pleomorphic sarcoma and a DL. There should be clear-cut evidence of ALT/WDL some distance from the dedifferentiated areas for the diagnosis of DL. Evaluating a high-grade sarcoma at its interface with normal fat results in an inappropriately low threshold for the diagnosis of DL. In cases where a clear preexisting component of ALT/WDL cannot be confidently distinguished from infiltration of fat by a spindle cell sarcoma, FISH for MDM2 can be helpful. Pleomorphic sarcomas showing MDM2 overexpression by IHC or amplification by FISH should be carefully examined for small, peripheral areas of ALT/WDL, as many such tumors will ultimately prove to represent DL, especially in the retroperitoneum.

A relatively rare problem is the significance of a microscopic focus of mitotically active, pleomorphic spindle cell sarcoma in an otherwise typical ALT/WDL. It has been suggested that dedifferentiation should be macroscopically visible (>1.0 cm) before the label “dedifferentiated liposarcoma” is applied, a reasonable proposal but one without supporting evidence at this time. From a practical perspective, we have often found that additional sampling of “borderline” lesions identifies larger areas of dedifferentiation, clarifying the diagnosis.