Nonlymphoid Hematopoietic Diseases Presenting in Bone, Soft Tissue, and Other Extranodal Sites: A Guide for the Everyday Surgical Pathologist Open Access

LCH is a clonal proliferation of Langerhans cells, which has previously been referred to by numerous names including histiocytosis X, eosinophilic granuloma, Hand-Schuller-Christian disease, Letterer-Siwe disease, and Langerhans cell granulomatosis. Descriptions of Langerhans cells and their ultrastructural features date back to the 1960s.^9,10  Soon thereafter, histiocytosis X was recognized as representing a proliferation of Langerhans cells.¹¹  Since that time, the understanding of LCH has grown exponentially through reported case series,^12–18  including numerous descriptions of LCH presenting at soft tissue sites.^12,19,20 

The disease is clinically heterogeneous and relatively uncommon. It is the most common of the dendritic cell neoplasms with an incidence of approximately 5 cases per 1 million population annually. Many cases present in childhood, with a slight male predilection. Presentation in the African American population is distinctly uncommon. Of note, primary LCH involving the lung is nearly always associated with cigarette smoking. Presentation can be unifocal, multifocal, or disseminated; the most common site of involvement by the unifocal form is bone (75%–80% of cases), with the skull, femur, vertebrae, and ribs representing the most common bony sites of involvement. Unifocal disease most commonly presents in older children and adults; the disseminated form is usually restricted to early childhood.²¹  Radiographically, the bone lesions are usually lytic and have well-circumscribed margins; however, sometimes they may have a permeative pattern associated with a periosteal reaction and mimic osteomyelitis and in children may mimic aggressive tumors such as Ewing sarcoma.

Recently, the genetic landscape of LCH has been explored, revealing mutations in the MAPK pathway, with a predominance of BRAF V600E mutations, which are seen in up to 50% of cases.^21–23  Approximately 30% of lesions have been shown to harbor immunoglobulin heavy chain (IGH), immunoglobulin κ chain (IGK), or T-cell receptor (TCR) rearrangements, with a few cases demonstrating concurrent B-cell and T-cell rearrangements.²¹ 

The prognosis in LCH is variable and depends on the disease burden, with nearly uniformly favorable outcomes in patients presenting with solitary lesions and mortality rates approaching 66% in the disseminated form with multivisceral involvement. Involvement of the bone marrow, liver, and lung are high-risk factors.^14,21 

In this example of LCH involving salivary gland, the tissue shows chronic sialadenitis (Figure 1, A, right) with an adjacent mixed inflammatory infiltrate (Figure 1, A, left). The inflammatory infiltrate is rich in eosinophils and contains many small, mature lymphoid cells, plasma cells, and numerous Langerhans cells with cleaved, “coffee-bean” nuclear contours and a modest amount of fluffy eosinophilic cytoplasm. (Figure 1, B).

By immunohistochemistry, the neoplastic Langerhans cells express CD68, CD1a (Figure 1, C), langerin (Figure 1, D), and S100 (Figure 1, D, inset).

These morphologic and immunohistochemical findings are classically seen in LCH. While the diffuse positivity seen for CD1a, langerin, and S100 in this case are typical, it is not uncommon for the positivity for any of these markers to be more variable.

The differential diagnosis of LCH commonly includes reactive processes, neoplasms rich in histiocytes, and other histiocytic neoplasms described later, including RDD, ECD, and ICH. The dense mixed inflammatory infiltrate present in LCH can be confused with a reactive process, including infectious processes such as osteomyelitis, autoimmune etiologies, and reactions secondary to trauma. A helpful feature is an abundance of associated eosinophils, which should prompt a search for intermingled clusters of Langerhans cells. These can be identified by their cleaved, “coffee-bean–like” nuclei. Immunohistochemistry is confirmatory as Langerhans cells stain positively with CD68, S100, CD1a, and langerin (CD207); however, one should be aware that the staining intensity of each stain may be variable and not all cells may stain. This immunoprofile also aids in the distinction of LCH from other neoplastic histiocytoses (Table 2). Similarly, other neoplasms rich in histiocytes, such as juvenile xanthogranuloma and reticulohistiocytoma, do not share the unique immunophenotype of LCH although they may enter the differential diagnosis. Of note, juvenile xanthogranuloma and reticulohistiocytoma often have numerous Touton-type giant cells, which are not common in LCH. Additionally, in challenging cases, positivity for BRAF V600E immunostain or molecular testing for BRAF mutations can be helpful in suggesting the presence of a clonal process such as LCH, although BRAF mutations have also been reported to a variable degree in the other histiocytoses discussed above.

Another entity in the differential diagnosis of a neoplasm with the same immunophenotype is Langerhans cell sarcoma (LCS). However, LCS is a frankly malignant neoplasm with high-grade cytologic features and brisk mitotic activity, which could otherwise be confused with malignant neoplasms of epithelial and mesenchymal differentiation on morphologic grounds alone. Thus, the morphologic distinction between LCH and LCS is usually straightforward, although the degree of atypia seen can be variable and cases with more mild degrees of atypia can be challenging. In general, the diagnosis of LCS should be reserved for cases that are overtly malignant and demonstrate immunohistochemical or ultrastructural evidence of Langerhans cell differentiation. It should be noted that LCS will generally not express all markers of Langerhans cell differentiation (S100, CD1a, langerin) and the expression of these markers may be very focal.

ICH is an exceptionally rare neoplasm—with approximately 100 cases reported in the literature to date—that arises from normal indeterminate cells, which are thought to be precursor cells to Langerhans cells. Patients present with multiple generalized skin lesions, described as papules, nodules, or plaques. No specific genetic abnormality has been described in ICH and the prognosis has been highly variable, ranging from spontaneous regression to rapidly progressive disease and death. Overall, little is known about this entity given its extreme rarity, although it was described several decades ago.^24–28  Of note, a few reports of patients with ICH responding to phototherapy have been reported.^29,30 

Histologic sections of a bone lesion show the classically described features of ICH, including involvement by sheets of cells with intermediate-sized nuclei, vesicular chromatin, and pink, fluffy cytoplasm (Figure 2, A). A higher-power image demonstrates that the nuclei are kidney-bean shaped and lack significant cytologic atypia (Figure 2, B). By immunohistochemistry, the neoplastic cells are positive for S100 (Figure 2, C) and CD1a (Figure 2, D) and negative for langerin (Figure 2, D, inset). The cells were also positive for CD68 (not shown).

The differential diagnosis of ICH is similar to that described for the previous histiocytoses. The diagnosis of ICH is not so much a morphologic diagnosis as it is an immunohistochemical diagnosis. While the presence of numerous histiocytes might raise the possibility of various entities, including LCH, RDD, and ECD, the presence of the characteristic CD68/S100/CD1a-positive and langerin-negative immunophenotype is necessary to secure the rare diagnosis of ICH.

ECD is a clonal histiocytic neoplasm, with fewer than 1000 bona fide cases described in the literature to date. ECD was originally described by Chester in 1930 as lipoid granulomatosis, which was later given the eponymous name of Erdheim-Chester disease in 1972.^21,31  ECD predominantly occurs in middle-aged patients and its diagnosis hinges largely on clinicoradiologic-pathologic correlation. While nearly any organ system can be involved by ECD, the bones are involved in most cases, commonly with a characteristic symmetrical osteosclerosis of the diaphyses and metaphyses of long bones. Cardiovascular involvement is characterized radiologically by a “sheathed” or “coated” aorta and renal involvement commonly produces a “hairy”-appearing kidney.^5,21,32–35 

The genetic alterations of ECD are seen in the MAPK pathway with BRAFV600E mutations being the most common. NRAS and PIK3CA mutations have also been described in this entity. ECD is a chronic, progressive disease with variable survival depending on the sites of disease involvement. Patients with central nervous system involvement or multivisceral disease commonly do poorly, while disease predominantly affecting the skeleton follows a more indolent course.^5,21,32  More recently, vemurafenib, a BRAF inhibitor, has shown promise in managing patients with ECD.³⁶  As more targeted therapies become available for the management of this disease, its prompt diagnosis is becoming critical.

Histologic sections of an unusual case of ECD involving juxta-articular tissue of the knee show a mixed inflammatory infiltrate including lymphocytes and plasma cells with numerous admixed foamy macrophages and scattered giant cells in an uncommon pseudopapillary-like architecture (Figure 3, A). In other areas of the lesion, sheets of histiocytes predominate with only scattered chronic inflammatory cells (Figure 3, B). At higher power, the mixed inflammatory infiltrate contains sheets of mononuclear cells with fluffy, eosinophilic cytoplasm (Figure 3, C). Careful histologic evaluation of histiocyte-predominant areas shows sheets of multivacuolated cells with variable amounts of eosinophilic cytoplasm (Figure 3, C, inset). At ×600 magnification, numerous small, mature lymphocytes and plasma cells amongst scattered histiocytes can be appreciated (Figure 3, D). Other than the papillary-like architecture seen in this case, the morphologic features, including mixed inflammatory infiltrate and sheets of histiocytes with a variable amount of fluffy, eosinophilic cytoplasm, are classic for ECD.

Immunohistochemical studies were not performed on this case; however, if they were to be performed, the sheets of cells with eosinophilic cytoplasm and foamy macrophages would be positive for CD68, CD163, and muramidase and negative for S100, CD1a, and langerin. A variable number of cases will be positive for BRAF V600E immunostain. The histologic and immunophenotypic features are rather nonspecific and can easily be misconstrued as reactive or even as other processes, such as tenosynovial giant cell tumor if juxta-articular in location, as in this case. Clinical and radiologic correlation is required in securing a diagnosis of ECD.

The differential diagnosis of ECD includes many of the same entities as LCH and RDD and navigation through this differential diagnosis is largely identical to that described above with some caveats. ECD commonly is associated with a mixed chronic inflammatory infiltrate and is usually not rich in eosinophils like LCH; however, some cases may be rich in eosinophils, complicating the differential diagnosis. Additionally, the presence of emperipolesis is notably lacking in ECD. Challenging small biopsy cases in which the presence of emperipolesis is uncertain can be more thoroughly assessed by S100 staining. S100 will stain only scattered dendritic cells, as the histiocytes in ECD are usually not S100 positive. Histiocytes in ECD commonly only express markers characteristic of normal histiocytes, such as CD68 and CD163, as described in Table 2.

The distinction of ECD from simply reactive chronic inflammatory infiltrates, particularly in bone lesions, can be exceedingly difficult, if not impossible, on histomorphology alone. Knowledge of the clinical and radiologic findings is necessary to make a definitive diagnosis of ECD, as certain imaging findings are characteristic, as previously described. Use of BRAF V600E immunohistochemistry can also be helpful in this regard, as many cases of ECD demonstrate this mutation. It should be noted, however, that this immunostain is neither entirely sensitive nor specific so knowledge of the clinical history and imaging findings is still highly advised.

RDD, previously designated sinus histiocytosis with massive lymphadenopathy, is a largely self-limited disorder, with an incidence of approximately 100 cases annually in the United States, that is likely neoplastic in etiology, although the underlying pathophysiology is not yet elucidated. The classic clinical presentation is that of massive, painless, bilateral cervical lymphadenopathy associated with fever and weight loss. It most commonly occurs in the first 2 decades of life, has a male predilection, and a propensity to affect African Americans. Involvement or presentation in extranodal soft tissues is not uncommon, occurring in 25% to 40% of patients, and the extranodal sites of involvement are usually skin, upper respiratory tract, soft tissue, orbit, bone, salivary gland, central nervous system, and breast.^14,37,38 

RDD typically has an indolent course, with most cases spontaneously resolving within a year, followed by complete remission. Rarely, the disease is associated with concurrent immunologic abnormalities and can be fatal.³⁷  Of note, we have seen rare cases of recurrent, clinically aggressive RDD with histologic transformation to histiocytic sarcoma. Therefore, while aggressive behavior should make one question the diagnosis, it does not exclude it.

RDD may not be at the top of most differential diagnoses when faced with a biopsy showing “nonspecific” inflammatory findings; thus, it is informative to review examples of RDD presenting in nonnodal locations.

Histologic sections of targeted core biopsies of a bone lesion show trabeculae of lamellar cancellous bone encased by a mixed inflammatory infiltrate (Figure 4, A) that replaces the hematopoietic marrow. At intermediate power (Figure 4, B), there are numerous large histiocytes with abundant pale foamy eosinophilic cytoplasm. While there is a suggestion of emperipolesis in some cells even at this power, it can be particularly challenging to identify in a limited and crushed sample. At higher power, the inflammatory infiltrate consists of histiocytes, plasma cells, neutrophils, and small, mature lymphoid cells with very few or no eosinophils (Figure 4, C). Some of the histiocytes are large and have abundant pale eosinophilic cytoplasm and round to oval vesicular nuclei with prominent nucleoli—these are the classic Rosai-Dorfman histiocytes. At high power, the presence of emperipolesis is better seen with phagocytized lymphocytes, plasma cells, and neutrophils; however, in challenging cases, emperipolesis can more easily be highlighted by S100 immunohistochemistry. Notice how the staining of the histiocyte cytoplasm is in sharp contrast to the negativity seen surrounding engulfed lymphoid cells (Figure 4, D). While the histiocytes are positive for CD68 and S100, they are negative for CD1a and langerin, in contrast to other lesions in the differential diagnosis (Table 2). The morphologic and immunohistochemical findings in this case are quite typical of those seen in biopsies of RDD involving extranodal sites. However, one should not expect to always see this degree of S100 positivity and obvious emperipolesis, particularly in small, crushed biopsies.

The differential diagnosis of RDD is very similar to that for LCH and includes reactive processes, osteomyelitis, neoplasms rich in histiocytes, and other histiocytic neoplasms. Distinguishing features include an absence of eosinophilia, lack of clusters of Langerhans cells, banal cytologic features, and the presence of large histiocytes with emperipolesis in RDD. Additionally, RDD has a distinct immunophenotype, demonstrating positivity for CD68, CD163, and S100 and negativity for CD1a and langerin (Table 2). OCT2 immunohistochemistry has also emerged as a marker for RDD, as it seems to be positive in most cases of RDD and negative in most cases of ECD and LCH.³⁹  Of note, emperipolesis is not specific for RDD and in small, crushed biopsies it may be difficult to appreciate, leading to a misdiagnosis; however, staining for S100 facilitates the identification of emperipolesis. In cases where sampling has not captured diagnostic RDD cells it can be difficult to distinguish from osteomyelitis. The distinction of RDD from other reactive and neoplastic processes is as discussed in the section on LCH.

Interestingly, we have seen cases of recurrent, locally aggressive RDD with the RDD cells acquiring malignant cytologic features, representing biological transformation to histiocytic sarcoma. While a rare occurrence, the presence of brisk mitotic activity or striking cytologic atypia should prompt careful evaluation, as the presence of atypia of questionable prognostic significance is well described in RDD and its distinction from progression to histiocytic sarcoma is not well defined.

Histiocytic sarcoma is a neoplasm composed of cytologically malignant cells that have the morphologic and immunohistochemical features of histiocytes and has largely become a diagnosis of exclusion made only after ruling out other neoplasms of histiocytic and other lineages; most reported cases are case reports or small series.^40,41  The exact incidence is nearly impossible to ascertain, as many lesions previously designated as histiocytic sarcoma have since been reclassified with the emergence of new immunohistochemical, cytogenetic, and molecular studies. The age at presentation is highly variable, ranging from infancy to the elderly, with most reported cases occurring in middle age. Presentation at extranodal sites is common with soft tissue and gastrointestinal tract involvement being the most commonly reported in bona fide cases.⁴²  Rare cases have been diagnosed in association with low-grade lymphomas, such as follicular lymphoma or chronic lymphocytic leukemia/small lymphocytic lymphoma. Another, perhaps unexpected, association is with mediastinal germ cell tumors.⁴³ 

No specific genetic abnormality has been described in histiocytic sarcoma, although identical genetic anomalies to those of the concurrent neoplasm in which they arise have been noted. Examples are the presence of the same i(12p) abnormality when arising in the setting of mediastinal germ cell tumors and exhibiting the same BCL-2 breakpoints as that present in associated follicular lymphomas. BRAF V600E mutations have also been variably reported in a subset of cases. In general, histiocytic sarcoma is an aggressive neoplasm with poor response to therapy; most patients succumb to progression of disease.²¹ 

Histologic sections of a classic example of histiocytic sarcoma show sheets of infiltrating cells with multiple areas of necrosis (Figure 5, A). The neoplastic cells are mostly polygonal, intermediate in size, and have vesicular nuclei, variably prominent nucleoli, and modest amounts of fluffy, eosinophilic cytoplasm. Some cells exhibit cytologic atypia in the form of nuclear membrane irregularities and hyperchromasia. Scattered neoplastic multinucleated osteoclast-like giant cells are also present (Figure 5, B). The neoplasm has an infiltrative growth pattern, splaying apart bundles of skeletal muscle (Figure 5, C). Areas of necrosis surrounded by neoplastic cells with striking cytologic atypia are present (Figure 5, D). The neoplastic cells are associated with a mixed inflammatory infiltrate rich in small, mature lymphocytes. While the presence of multinucleated osteoclast-like giant cells and tumor necrosis are seen in this classic example, these features are not always seen. Some cases, particularly small biopsy samples, may not have demonstrable infiltration, necrosis, or marked nuclear atypia. In cases lacking these features and having only limited nuclear atypia, a conservative diagnosis of “atypical histiocytic proliferation” is acceptable, as the degree of atypia in histiocytic neoplasms may be variable and a specific cutoff for malignancy is not currently established. In these cases, a conservative excision can be recommended for further evaluation.

The neoplastic cells are positive only for CD163 (Figure 5, A, inset) and CD68 (not pictured) and negative for markers characteristic of other lineages, including multiple cytokeratins, smooth muscle actin (SMA), desmin, S100, CD1a, langerin, CD21, CD23, and CD35. The Ki-67 nuclear proliferation index is brisk (∼40%).

The differential diagnosis of histiocytic sarcoma includes a wide array of poorly differentiated neoplasms of various lineages. As histiocytic sarcoma is largely a diagnosis of exclusion, poorly differentiated neoplasms of any lineage should be excluded before making the diagnosis. As such, a broad panel of markers directed at other cell types (ie, pancytokeratin, S100, B- and T-cell markers, and mesenchymal markers) should be applied before rendering a definitive diagnosis.

The most common differential diagnoses that arise in this setting are other histiocytic/dendritic cell neoplasms, such as FDCS and IDCS. While the background inflammatory infiltrate rich in small, mature lymphoid cells can be seen in all of these entities, FDCS and IDCS commonly consist of short fascicles of spindle cells rather than polygonal cells, as are seen in histiocytic sarcoma. Additionally, the immunohistochemical profiles of FDCS and IDCS distinguish them from histiocytic sarcoma (Table 3).

FDCS is an entity with an unknown exact incidence and reported cases in the literature numbering only in the hundreds, which was previously referred to as dendritic reticulum cell tumor,^44,45  the normal counterpart of which is the follicular dendritic cell. FDCS commonly presents in middle age with no documented sex predilection. Patients often present with a slow-growing mass, which is often large at diagnosis. Approximately one-third of cases are nodal and about 60% of cases are both nodal and extranodal, with the few remaining cases being solely extranodal. The most affected extranodal sites are the tonsil, gastrointestinal tract, soft tissue, mediastinum, retroperitoneum, and lung.^21,46–51  Of note, FDCS has been frequently reported in association with the hyaline vascular variant of Castleman disease.

While no specific genetic abnormalities have been described in FDCS, recent studies have shown a tumor suppressor–driven biology with mutations in various genes including CDKN2A, RB1, BIRC3, and CYLD.⁵²  Additional studies have shown a tumor microenvironment enriched in T follicular helper cells and T regulatory cells with increased levels of PD1.⁵³  In the same vein, anecdotal cases of FDCS have been shown to respond to immune checkpoint therapy.⁵⁴ 

In general, FDCS is a relatively aggressive disease, with local recurrence and distant metastatic rates approaching 30%. Commonly, recurrence or metastasis develops several years after surgical resection. Large tumor size, coagulative necrosis, high mitotic count, and significant cytologic atypia are associated with more aggressive disease.⁵⁵ 

Histologic sections of core biopsies show involvement by sheets of spindle cells with eosinophilic cytoplasm (Figure 6, A). A higher-power photomicrograph demonstrates that the tumor cells are spindled in morphology with variable nuclear atypia, hyperchromatic nuclei, and a moderate amount of eosinophilic cytoplasm (Figure 6, B). In areas of less cellular growth, one can appreciate the admixed inflammatory cells, with a predominance of small, mature lymphoid cells and scattered neutrophils and eosinophils (Figure 6, C). These morphologic findings are highly characteristic of the findings expected in FDCS.

By immunohistochemistry, the neoplastic cells are positive for CD21 (Figure 6, D) and CD35 (Figure 6, D, inset). The tumor cells would also be variably positive for CD23, epithelial membrane antigen (EMA), clusterin, and D2-40 and negative for S100; however, these images are not available for review. While this is the classic expression profile of FDCS, all of these markers may not display positivity in any given case and the degree of positivity is usually more variable, unlike the uniform positivity seen in this example.

The differential diagnosis of FDCS most commonly involves other dendritic cell neoplasms; however, if one is not well versed in the histomorphologic appearance of FDCS, limited biopsy specimens may mimic several other mesenchymal neoplasms. A helpful hint that a dendritic cell neoplasm might be present is the intimate admixture of lymphoid cells associated with the neoplastic spindle cell proliferation. FDCS is further distinguished from other dendritic cell neoplasms by its immunophenotype, as demonstrated in Table 3. FDCS will be positive for CD21, CD23, and CD35, albeit variably and all markers may not be expressed, and negative for S100; this contrasts with immunophenotypes of the morphologically similar IDCS and FRCT, discussed below. Additionally, FDCS will often demonstrate positivity for EMA, clusterin, and D2-40, further aiding its distinction from lesions in the differential diagnosis. The distinction between FDCS and neoplasms of mesenchymal origin will become apparent when the lymphocytic infiltrate is appreciated and mesenchymal markers display negativity. A diagnosis of a dendritic cell neoplasm, such as FDCS, should then be entertained in the setting of a spindle cell neoplasm lacking other lines of differentiation.

IDCS is a dendritic cell neoplasm that has only been described in case reports and small case series totaling approximately 100 reported cases.^56–60  While lymph node involvement is the most common presentation, involvement of extranodal sites, including the skin, stomach, adrenal gland, and mesentery, has also been documented.⁶¹  Most cases are asymptomatic, although presentation varies with the area of involvement and mass size.

Like many of the neoplasms discussed, IDCS has been noted to arise in association with other hematopoietic neoplasms such as low-grade B-cell lymphoma. In cases associated with follicular lymphoma, identical BCL-2 breakpoints have been identified in the follicular lymphoma and associated IDCS. No specific genetic alterations have been associated with IDCS to date, although few data are available on its genetic underpinnings.⁶² 

In general, IDCS is an aggressive disease, with most patients dying of disease. Higher stage at diagnosis correlates with poor clinical outcome.^21,59,63 

Histologic sections of a typical case of IDCS involving the parotid show parotid parenchyma involved by a relatively well-circumscribed spindle cell neoplasm. Even at low power, a storiform growth pattern of neoplastic spindle cells can be appreciated (Figure 7, A). Higher magnification shows that the neoplastic spindle cells are variable in size with open, vesicular chromatin and variably prominent nucleoli. The cells are cytologically atypical and have a moderate amount of pink cytoplasm (Figure 7, B). Notice scattered foci within the neoplasm that are richly infiltrated by small, mature lymphoid cells with fewer plasma cells, neutrophils, and eosinophils, as is typical of dendritic cell neoplasms (Figure 7, C). In addition to the nuclear atypia, mitoses are also easily appreciated (Figure 7, D). While this example case demonstrates obvious nuclear atypia and numerous mitoses, these features are oftentimes more subtle in IDCS.

The neoplastic cells in this case were strongly and diffusely positive for S100 and negative for CD21, CD23, CD35, and SOX10, as discussed in Table 3. It should be noted that while S100 positivity is required in making this diagnosis, the degree of positivity is usually patchy rather than diffuse, as seen in this case. This is particularly true for more poorly differentiated/higher-grade tumors.

The differential diagnosis of IDCS, like FDCS, includes other spindle cell neoplasms of mesenchymal origin and FDCS. Resolution of this differential diagnosis rests on the pathologist's ability to consider dendritic cell neoplasms when encountered with a spindle cell neoplasm, particularly in a lymph node. Once thought of, the diagnosis is made relatively easily with immunohistochemical stains. The staining profile of IDCS is S100 positive and CD21/CD23/CD35 negative, as shown in Table 3. While the S100 positivity may raise the differential diagnoses of metastatic spindle cell melanoma or malignant peripheral nerve sheath tumor (MPNST), negativity for SOX10, as seen in this case, may aid in resolving this differential; however, it should be noted that IDCS can demonstrate positive staining for SOX10. In these instances, positivity for CD45 and CD68 seen in IDCS will be helpful. In challenging cases, close correlation with clinical history and even molecular studies may be required to suggest the appropriate diagnosis. As SOX10 immunostaining is commonly negative or weak and focal in MPNSTs of spindle cell morphology, appreciating that the tumor has a prominent lymphocytic infiltrate would be most helpful in recognizing that MPNST is less likely. MPNSTs are also likely to have more severe cytologic atypia and prominent mitotic activity. Additionally, a subset of MPNSTs will demonstrate loss of H3K27me3 by immunohistochemistry, which can be helpful in challenging cases. Given the relative rarity of IDCS, each of these differential diagnoses should be carefully considered and extensively ruled out before rendering a diagnosis of IDCS.

FRCT is an exceedingly rare entity with fewer than 50 cases reported in the literature to date. These tumors have also been referred to as cytokeratin-positive interstitial reticulum cell tumors and have been documented to present in lymph nodes, spleen, and soft tissue.^64–68  Not enough data are available to assess the genetic landscape or overall prognosis of this neoplasm, although the few reported cases had locally aggressive growth akin to that of a low-grade sarcoma.

This entity is sufficiently rare that we do not currently have any hematoxylin-eosin examples in our archives. As such, we will discuss the expected findings below to ensure that our colleagues are at least aware of its existence.

Histologic sections of FRCT generally show features akin to those seen in FDCS or IDCS, including the presence of blunt spindle cells arranged in short fascicles and whorls with a moderate degree of cytologic atypia, nuclei with vesicular chromatin, variably prominent nucleoli, and a moderate amount of eosinophilic cytoplasm. As with the other dendritic cell tumors described above, the presence of intimately admixed inflammatory cells can be a clue to the lineage of this neoplasm.

The main distinction between FRCT and FDCS/IDCS is thus not morphologic but immunophenotypic, as FRCT will be variably positive for CD68, SMA, desmin, and pancytokeratin and negative for S100, CD21, CD23, and CD35, unlike the aforementioned differential diagnoses. The pancytokeratin positivity will be in a dendritic cell pattern, with multiple cellular processes staining, as shown in Figure 8.

The differential diagnosis of FRCT includes both FDCS and IDCS, as well as other spindle cell mesenchymal neoplasms. The myofibroblastic differentiation seen in this neoplasm can be helpful in suspecting the possibility of its presence, particularly in lymph node biopsies, if appreciated on routine hematoxylin-eosin staining. The presence of variable SMA, desmin, cytokeratin (in a dendritic cell pattern), and CD68 immunostaining and negativity for S100, CD21, CD23, and CD35, as described in Table 2, are suggestive of the diagnosis. However, other neoplasms of myofibroblastic and smooth muscle differentiation must be ruled out before considering this diagnosis. This requires use of additional immunohistochemical stains, including desmin and other smooth muscle markers to rule out metastasis of leiomyosarcoma. While FRCT is commonly variably positive for these markers, it does not demonstrate the degree of staining as is usually seen in leiomyosarcoma. Additionally, metastatic leiomyosarcoma typically has a higher degree of cytologic atypia along with other high-grade features (increased mitotic activity, necrosis). Inflammatory myofibroblastic tumor (IMT) is another diagnostic consideration in these cases. Helpful features favoring a diagnosis of FRCT include the often dim desmin and SMA positivity, as staining for desmin and SMA is often strongly positive in IMT; the positivity of FRCT for CD68; and the characteristic dendritic cell pattern of staining of FRCT for cytokeratin. Additionally, although not positive in all cases of IMT, ALK positivity would support a diagnosis of IMT over FRCT. High-grade myofibroblastic/fibroblastic sarcoma (undifferentiated pleomorphic sarcoma) might also enter the differential diagnosis given the positivity for SMA and negativity for other markers. FRCT will also often express at least focal CD68, a helpful differentiating characteristic. Additionally, undifferentiated pleomorphic sarcoma often has high-grade features that are usually lacking in FRCT.

Acute myeloid leukemia presenting as solitary bone and soft tissue lesions has long been recognized by various names, including granulocytic sarcoma, myeloid sarcoma, chloroma, and extramedullary myeloid tumor.²¹  The most frequent extramedullary sites affected are the skin, gastrointestinal tract, bone, soft tissue, and testes. While many patients presenting with myeloid sarcoma have an antecedent history of myeloid neoplasm such as myelodysplastic syndrome or acute leukemia, a subset will not have this history, making recognition of these neoplasms more difficult, potentially leading to improper and/or delayed therapy.^69–71  The incidence of granulocytic sarcoma presenting in the absence of an underlying myeloid neoplasm has been estimated at 2/1,000,000.⁷²  Interestingly, certain genetic aberrations are associated with increased frequency of soft tissue involvement in myeloid neoplasms. These include acute myeloid leukemia with NPM1 mutations; core-binding factor leukemias, including t(8;21) and t(16;16)/inv(16) acute myeloid leukemias; and MLL-rearranged acute myeloid leukemia. In general, myeloid sarcoma has a poor prognosis, which may be improved by bone marrow transplant, which improves overall survival up to 47% in some studies.²¹ 

Core biopsy of an omental mass in a patient with myeloid sarcoma shows a diffuse cellular infiltrate of predominantly intermediate-sized cells with open, powdery chromatin and variably prominent nucleoli arranged in discohesive sheets and invading fibroadipose tissue (Figure 9, A through D). This morphology, although not specific, is very characteristic of myeloid sarcoma.

By immunohistochemistry, the neoplastic cells are positive for CD43, CD34, CD117, myeloperoxidase (MPO), and muramidase (Figure 10, A through D). Additional immunohistochemical stains for pancytokeratin, CD19, CD20, CD3, and S100 displayed negativity. While this case demonstrates positivity for CD34 and CD117, positivity for these markers in myeloid sarcoma is commonly diminished or even completely absent. Even in the absence of these markers, the pathologist should feel comfortable making the diagnosis of myeloid sarcoma, based on the overall morphology and immunophenotype, including infiltration by blastic cells expressing a hematopoietic marker such as CD43 and CD45 and lacking markers of other lineages. Of note, the expression of CD45 and CD43 in these neoplasms is also commonly weak and focal.

As the histologic appearance is that of a poorly differentiated malignant neoplasm, the differential diagnosis on morphologic grounds alone is quite broad and includes poorly differentiated carcinoma, round cell sarcoma, and malignant melanoma, as well as numerous hematopoietic neoplasms. The discohesive nature of the cells and blastic appearance of the chromatin can be clues that one is dealing with a hematopoietic neoplasm, although poorly differentiated carcinomas and melanoma can mimic this morphology. A limited initial immunohistochemical panel that would be helpful in narrowing the differential diagnosis includes CD43, pancytokeratin, and S100. In our experience, CD43 is a superior screening marker for high-grade hematopoietic neoplasms, as compared to CD45. Positive CD43 immunohistochemical staining in the presence of negative pancytokeratin and S100 staining in a case with this morphology would support that the neoplasm is hematopoietic in nature. Lesions that may enter the differential diagnosis include acute leukemia and high-grade B- and T-cell lymphomas. Additional immunohistochemical stains, including CD34, TdT, CD117, and lineage-defining markers (ie, MPO, CD19, CD3, muramidase) would be required to narrow the differential diagnosis and arrive at a more precise diagnosis. As with many of the neoplasms described in this article, the key to making the correct diagnosis is being aware of this entity. Once the diagnosis is entertained, immunohistochemistry can usually easily confirm it.

BPDCN is a hematopoietic tumor derived from precursors of plasmacytoid dendritic cells considered to be of myeloid lineage. These neoplasms have an annual incidence of slightly more than 1000 cases in the United States and Europe combined and arise mainly in elderly patients but have been described in all age groups. Presentation in the skin and underlying subcutaneous tissue is the most common manifestation of disease, occurring in 60% to 100% of cases; bone marrow and peripheral blood involvement are also common at presentation. The cutaneous manifestation of disease is classically referred to as “a bruise that won't heal,” and can be quite dramatic clinically.²¹  To date, the prognosis for patients diagnosed with BPDCN is quite dismal, with a median survival of 10.0 to 19.8 months, and therapy options have been limited.^73–76  However, as the molecular genetics of this neoplasm are unraveled, new targeted therapies have been discovered that may improve patient outcomes. These include targeting of interleukin-3 receptor subunit α (IL3RA or CD123), which is expressed in virtually all cases of BPDCN, and inhibition of the NF-kB and TCF4/BRD4 pathways, which are upregulated in this entity.⁷⁷  Although most BPDCNs have karyotypic abnormalities, they are not disease specific and include mostly deletion abnormalities involved in cell cycle regulation. MYC rearrangements leading to overexpression of MYC protein have been reported in 38% of BPDCN cases. BPDCN has a mutational profile overlapping with that of other myeloid malignancies.²¹ 

Histologic sections of skin in a typical case of BPDCN show sheets of primitive-appearing cells infiltrating the superficial and deep dermis and subcutaneous adipose tissue. The tumor cells are intermediate in size with open, powdery chromatin and a minimal amount of cytoplasm (Figure 11, A and B). Note the abundant intralesional extravasation of red blood cells (Figure 11, B).

By immunohistochemistry, the neoplastic cells are positive for CD4 (Figure 11, C) and CD56 (Figure 11, D) and negative for CD3 (Figure 11, C, inset). The coexpression of CD4 and CD56 in the absence of other T-cell markers in this clinical setting is highly suggestive of BPDCN. The neoplastic cells are also positive for CD68 in a characteristic “dotlike” pattern (Figure 12, A) and demonstrate variable positivity for TdT (Figure 12, B), while being negative for lineage-defining blast markers including CD19 (Figure 12, C), MPO (Figure 12, D), and muramidase (not shown). Of note, TCL-1 immunostaining is also commonly positive in BPDCN and can be useful in the differential diagnosis; it was not performed in this case. The demonstrated immunophenotype is classic for BPDCN; however, it should be noted that the characteristic “dotlike” pattern of CD68 is not always observed. Additionally, the relatively strong and uniform expression of both CD4 and CD56 seen in this case is not always observed, as these markers more commonly display patchy and variable positivity.

One of the most helpful clues to the diagnosis of BPDCN is the presence of a blastic neoplasm expressing CD4 without expression of other T-cell markers. The differential diagnosis of BPDCN is akin to that of acute leukemia presenting as an extramedullary mass, as the nuclear size and appearance of the chromatin are very similar. In addition to the differential diagnoses discussed in the section on acute leukemia, Merkel cell carcinoma may also enter the differential diagnosis of BPDCN, particularly when presenting in the sun-exposed skin of an older individual. An initial immunopanel as discussed in the previous section would be appropriate, possibly with the inclusion of CK20 depending on level of suspicion for Merkel cell carcinoma. When the initial immunopanel suggests the hematopoietic nature of the neoplasm, the addition of lineage-defining markers will help in narrowing the differential diagnosis. When BPDCN is suspected, CD123, muramidase, CD68, CD163, CD56, and CD4 should specifically be added to the immunopanel. TCL-1 immunostaining can also be useful, as it is commonly positive in BPDCN and negative in potential mimics.

BPDCN not uncommonly demonstrates variable TdT staining, which can raise concern for lymphoblastic leukemia/lymphoma; however, the absence of B- and T-cell lineage-defining markers is evidence against this possibility. It is common for the main differential diagnoses to include myeloid sarcoma associated with an acute leukemia with monocytic differentiation and BPDCN. In fact, this differential can be quite challenging if not impossible to distinguish, based on morphologic and immunohistochemical findings alone, particularly when multiple stains show only focal and/or weak staining. Correlation with clinical findings and cytogenetic and molecular data are paramount in making this distinction, as the presence of a recurrent cytogenetic abnormality seen in AML, such as MLL rearrangement for instance, would lead to a diagnosis of AML. In the absence of these abnormalities, distinction can remain difficult without gene expression analysis; however, the unique gene expression profile seen in BPDCN currently supports its inclusion in a category of its own. Oftentimes, a diagnosis of blastic hematopoietic neoplasm involving the skin with the provided differential diagnosis will suffice in biopsy material to guide clinical management.

Although many of the entities discussed above are uncommon, particularly when presenting at extranodal sites, they should always be considered in the differential diagnosis when one is confronted with a neoplasm of uncertain lineage. More specifically, one should remember that when confronted with a “mass lesion” and histologic sections only reveal what appears to be a mixed inflammatory infiltrate, entities of histiocytic origin should be considered and the appropriate stains, including CD68, S100, CD1a, and langerin, should be ordered to explore these possibilities. When a spindle cell neoplasm with brisk associated chronic inflammatory infiltrate is seen, neoplasms of dendritic cell origin should at least be considered as part of the differential and appropriate stains including S100, CD21, CD23, CD35, EMA, clusterin, and D2-40 should be applied, particularly when neoplasms of mesenchymal, melanocytic, and epithelial origin have been excluded by a battery of immunostains including numerous keratins, p63, p40, SMA, desmin, CD34, and SOX10. Neoplasms with high-grade/blastic morphology presenting in soft tissue should be evaluated for the possibility of myeloid sarcoma or BPDCN, especially when all other lineage-defining markers display negativity. This can be done by the application of a panel including CD34, CD117, MPO, muramidase, CD4, CD3, CD19, and CD56. In poorly differentiated neoplasms with no morphologic evidence of line of differentiation, an initial screening panel including pankeratin, S100, and CD43 is prudent. Negativity for pankeratin and positivity for CD43 in this setting should prompt the pathologist to consider the malignant entities discussed above, including myeloid sarcoma, BPDCN, histiocytic sarcoma, among others.

In conclusion, it is not uncommon in our consult practice to receive examples of these entities in which all other lineages have been explored and excluded other than histiocytic/dendritic/myeloid cell lineage. This tells us that they are uncommonly considered in many pathologists' differential diagnoses. It is our hope that this review of these entities aids the everyday surgical pathologist in their consideration and recognition.