by Dr Weijie Li
Cytogenetic and molecular testing has been becoming more and more important in the diagnosis of human diseases. These tests are indispensable for the diagnosis and/or prognostic prediction for some hematopoietic malignancies. This article discusses a group of lymphomas with similar morphology and phenotype but distinct cytogenetic/molecular abnormalities, with emphasis on the importance of proper cytogenetic/molecular testing in the diagnosis of these diseases.

Background

Burkitt lymphoma (BL) is an aggressive high-grade B-cell lymphoma (HGBCL), which typically shows sheets of monomorphic mediumsized tumour cells with round nuclei, finely dispersed chromatin, and multiple basophilic small to intermediate-sized nucleoli (Fig. 1a). The cytoplasm is moderately abundant and highly basophilic with multiple lipid vacuoles better visualized on Wright’s and/or Giemsa stained air-dried smears or imprint slides. There are many mitotic and apoptotic figures, and numerous intermixed tingible body macrophages resulting in a so-called ‘starry sky’ pattern. The immunophenotype of typical BL is that of germinal centre B-cell-type (GCB), positive for immunoglobulin (Ig) M, CD19, CD20, CD22, and CD79a, CD10 and BCL6 (see Table 1 for definitions). BCL2 is usually negative. The proliferative index (as determined by Ki-67 levels) is nearly 100% (Fig. 1a) [1].
There are three clinical variants of BL: endemic, sporadic and immunodeficiency-associated. They differ mainly in their geographic distribution, clinical presentation and anatomic localization of primary tumour. Regardless of the clinical variants, the molecular hallmark of BL is the translocation of the MYC proto-oncogene (MYC) (8q24) to the Ig heavy chain region (14q32), t(8;14) (q24;q32), or less commonly to the Ig light chain kappa locus on 2p12 or the Ig light chain lambda locus on 22q11. There are other molecular cytogenetic changes, which include copy-number gains involving 1q, 7 and 12 and losses involving 6q, 13q32-34 and 17p. Gene expression profiling studies have defined a molecular signature characteristic of BL, which is different from other lymphomas such as diffuse large B-cell lymphoma (DLBCL) [2]. Mutations of the transcription factor 3 (TCF3) gene or the gene for its negative regulator, inhibitor of DNA binding 3 (ID3) have been frequently detected by next-generation sequencing analysis in sporadic BL cases. The resulting mutant proteins can activate B-cell receptor signalling, which sustains BL cell survival by engaging the phosphoinositide- 3-kinase pathway. Other mutations in CCND3, TP53, RHOA, SMARCA4 and ARID1A genes are also detected in certain BLs. Endemic cases show fewer mutations overall and lower proportion of mutations in TCF3 or ID3 [1].
Technically, all BL cases should have MYC rearrangement. It is controversial to make a diagnosis of BL when MYC rearrangement is not detectable. However, routine fluorescence in situ hybridization (FISH) and chromosomal analysis may not be able to detect this translocation owing to complex karyotype or more complicated translocation. Additionally, by gene expression profile study, some MYC-negative cases show a typical BL molecular signature [2]. Therefore, in the presence of classic morphology and phenotype, the diagnosis of BL should be made even without the demonstration of this translocation. Recent studies [3, 4] show that some true MYC-negative cases have characteristic 11q aberration, which has been recognized by the 2016 WHO classification as a provisional new entity [1, 5] and will be discussed more in detail next.
Historically, there has been no consensus on the diagnostic classification of the lymphomas with morphology resembling BL but showing more cytological pleomorphism and/or other atypical morphologic or phenotypic features. These were so-called ‘grey-zone lymphomas’ and classified as atypical BL/Burkitt-like lymphoma (BLL) by the 2001 WHO Classification of tumours of hematopoietic and lymphoid tissues. The name was then changed to B-cell lymphoma, unclassifiable, with features intermediate between BL and DLBCL by 2008 WHO classification [6]. These lymphomas were thought to represent a continuum between BL and DLBCL.

2016 revision of the WHO classification

The 2016 WHO classification of tumours of hematopoietic and lymphoid tissues has shed some light on this field with the modification of the grey-zone lymphoma with features intermediate between BL and DLBCL, and the creation of Burkitt-like lymphoma with 11q aberration (BLL-11q) and large B-cell lymphoma with IRF4 rearrangement (LBL-IRF4) [1, 5].
HGBCL with MYC and BCL2 and/or BCL6 rearrangement and HGBCL, not otherwise specified
The grey-zone lymphoma with features intermediate between BL and DLBCL has been divided into two categories: HGBCL with MYC and BCL2 and/or BCL6 rearrangement (double-hit or triple-hit lymphomas); and HGBCL, not otherwise specified (NOS). These lymphomas usually occur in older adults, with men and women affected equally. They are usually present as a rapidly enlarging mass involving lymph nodes or extranodal sites with advanced stage. These lymphomas have very poor prognosis with no optimal therapeutic strategy [7–9]. Double-hit (DH)/triple-hit (TH) refers to the co-occurrence of MYC and BCL2 and/or BCL6 translocations. MYC rearrangements usually differ from those seen in BL: they often involve non-IG partner genes. These lymphomas usually have complex karyotypes. The majority (60–70%) of them have MYC/BCL2 rearrangements, with a minority (15–30%) harbouring MYC/BCL6 rearrangements and fewer (10%) harbouring rearrangements of all three genes. Rare cases of follicular lymphoma or B-lymphoblastic leukemia/lymphoma can have these DH/TH cytogenetic changes. They should not be included in this category. HGBCLs with DH/TH morphologically can resemble BL, DLBCL or lymphoblastic lymphoma. They are phenotypically mature B-cell lymphomas with the presence of CD19, CD20, CD79a, and Pax-5 and lack of TdT proteins. Most cases also produce CD10 and BCL6. HGBCL, NOS includes cases with atypical BL features or blastoid morphology without DH or TH cytogenetic findings, regardless of MYC status. They are phenotypically mature B-cell lymphomas. For cases with blastoid morphology, cyclin D1 and SOX-11 should be stained for to rule out blastoid variant of mantle cell lymphoma. Approximately 20–35% of HGBCL, NOS cases have a MYC rearrangement [1].
BLL-11q lymphomas
BLL-11q encompasses cases with morphologic, phenotypic, and gene expression resemblance to BL, but lacking MYC translocation and harbouring characteristic proximal 11q gains at 11q23.3 and distal 11q loss at 11q24-qter (Fig. 1b) [1, 3–5]. These 11q changes are likely to be the result in the upregulation of oncogenes PAFAH1B2, USP2, and CBL located in the gained regions of 11q23, and corresponding down-regulation of tumour suppressor candidate genes FLI1, ETS1, TBRG1, and EI24 located in the regions of 11q24-qter loss. Cases of BLL-11q commonly show some degree of cytomorphologic pleomorphism and tend to have a more complex karyotype than BL [1]. The phenotype is that of GCB in nearly all cases [3, 4]. Besides the classic gains and losses within 11q, additional frequently identified abnormalities include del(6q) and trisomy 12 [10]. A recent study has shown that BLL-11q lymphomas have mutational landscape distinct from BL [11], which indicates that BLL-11q is truly a distinct entity. BLL-11q cases occur over a wide age range but are more common in children and young adults. They are more frequently nodal than BL and tend to present as a single dominant mass or conglomerate mass [4, 10]. Patients tend to present with limited disease without involvement of bone marrow or central nervous system, and prognosis appears to be favourable, similar to classical BL [1].
LBL-IRF4
LBL-IRF4 most commonly affects children and young adults [1, 5]. It mainly involves the Waldeyer’s ring and cervical lymph nodes and usually presents as low stage disease. Microscopically, the tumour cells are medium to large with finely clumped chromatin and small basophilic nucleoli. A starry sky pattern is usually absent, though proliferation rate is usually high by Ki-67 stain. These lymphomas may have a diffuse growth pattern, follicular growth pattern, or follicular/diffuse pattern. The tumour cells are positive for B-cell specific markers (CD20, CD79a, Pax-5), and characteristically show high levels of IRF4and BCL6. Over 50% of the cases are also positive for BCL2 and CD10. Despite the high levels of IRF4, these cases have a GCB signature by gene expression profiling. Most cases have a cytogenetically cryptic rearrangement of IRF4 with an Ig heavy chain locus. BCL6 alterations may be detected in some cases, but essentially all cases lack MYC and BCL2 rearrangement. Most cases have shown good response to chemotherapy [12, 13].

Diagnosis of these lymphomas

All the lymphomas described above except BL are uncommon and quite confusing to most general pathologists. They can be easily misdiagnosed in the absence of a proper work-up plan. Based on our practical experience, a step-by-step diagnostic algorithm has been designed, as shown in Figure 2. This diagnostic scheme is not recommended for the practice in BL endemic areas, where morphology and phenotype are likely to be adequate for the diagnosis of most BL cases. When a lymphoid malignancy is encountered with BL or BLL morphology, and mature B-cell phenotype is demonstrated by flow cytometry and/or immunochemical staining, the first check required is for MYC rearrangement. If MYC rearrangement is detected by FISH or other methods, BL will be the diagnosis in the presence of classic BL morphology and phenotype. If there is morphologic and/or phenotypic atypia, further testing for BCL2 and BCL6 rearrangement is necessary. In the presence of BCL2 and/or BCL6 rearrangement, the diagnosis will be HGBCL with DH or TH. Without the rearrangements of BCL2 and BCL6, the diagnosis will be HGBCL, NOS. Based on our experience in a children’s hospital in the USA, sporadic BL cases frequently show a certain degree of atypical morphologic features. And HGBCL with DH or TH is exceedingly rare in children [7, 8, 14]. Therefore, testing for DH/TH is usually not necessary for pediatric cases. These pediatric cases should be diagnosed as BL as long as MYC rearrangement is detected unless morphologic and phenotypic atypia is significant. The HGBCL, NOS category is not applicable for pediatric cases, considering the excellent prognosis of most pediatric HGBCL cases and their distinct molecular features [14–16]. In the pediatric population, MYC+ cases with significant morphologic and/or phenotypic atypia but no DT/TH should be diagnosed as DLBCL, NOS.
Although high Myc levels (in >80% of nuclei) are present in most cases of BL, there is much more variation in the HGBCLs with DH/ TH. Although most studies have concluded that Myc staining is not reliable enough to be used for the selection of cases for cytogenetic or molecular testing, some studies suggest using a cut-off of >30% or >70% Myc positivity for case selection [17, 18].
The rearrangements of MYC, BCL2, and BCL6 should be detected by a cytogenetic/molecular method such as FISH. The presence of only copy-number changes or somatic mutations, without an underlying rearrangement, is not enough to qualify a case for this category. So-called double-expresser DLBCLs show immunohistochemical overexpression of Myc and BCL2 protein and have a relatively poor prognosis [19]. However, overexpression cannot be used as a surrogate marker for DH cytogenetic status, because most double-expressers are not DH lymphomas (although most DH lymphomas are also double-expressers).
If MYC rearrangement is absent (MYC−), the newly proposed entity, BLL with 11q, should be considered and properly tested. The diagnosis of BLL-11q is based on the presence of characteristic gain/loss patterns of 11q, together with BL/BLL morphology, GCB phenotype, and lack of MYC rearrangement. The characteristic 11q aberration is key to making the diagnosis, but its presence alone is neither specific nor diagnostic since it may also be present in MYC+ BL or DLBCL [10, 20]. The most sensitive method for detecting this characteristic cytogenetic finding is DNA microarray. The 11q aberration can be visualized by chromosomal analysis. However, chromosomal analysis relies on tumour cell viability and metaphase morphology, and the finding may not be characteristic for this aberration if the resolution is low. Another potential diagnostic strategy is FISH for chromosome 11 abnormalities. Commercially available FISH probes for chromosome 11 regions may be used to detect gains within 11q and 11q terminal loss. As a result of the variation of gain/loss spots among the cases, depending on the probes used, FISH alone may miss some cases. If MYC rearrangement is absent and there is no typical 11q aberration, the diagnosis should be HGBCL, NOS for adult patients. For pediatric patients, these lymphomas often have a Burkitt or intermediate molecular gene expression profile and show an excellent prognosis. These cases should not be classified as HGBCL, NOS. For the pediatric cases with morphology and phenotype close to BL and relatively simple karyotype, the diagnosis should be BL; otherwise the diagnosis should be DLBCL, NOS. If the case shows diffuse and levels of of IRF4 and BCL6, especially in Waldeyer’s ring and cervical regions, LBL-IRF4 with diffuse growth pattern should be considered. The diagnosis of LBL-IRF4 should be confirmed by FISH analysis for IRF4 rearrangement [21].

Conclusion

With the advances in cytogenetic/molecular studies of lymphomas with BL or BLL morphology, new classification of these lymphomas has been proposed. Accurate diagnosis of these lymphomas needs the combination of cytogenetic/molecular testing results, morphology and phenotype. More specific or targeted treatment for these lymphomas may be on the horizon, and hence accurately diagnosing them is clinically important.

The author

Weijie Li MD, PhD Department of Pathology and Laboratory Medicine,
Children’s Mercy Hospital, University of Missouri-Kansas City School of
Medicine, Kansas City, MO 64108, USA

E-mail: wli@cmh.edu

by Dr J. Filipe, Dr D. V. Pereira and Dr S. André
Granular cell tumours of the breast are rare, mostly benign neoplasms derived from Schwann cells. Clinical and radiological features may be worrisome, raising the possibility of malignancy. The only method for diagnosis is histopathological evaluation of the lesion, mandatory for the appropriate management of patients.

Background

Granular cell tumours (GCTs) were first described in the tongue in 1926 by the Russian pathologist Abrikossoff, who suggested a myofibroblastic origin [1]. He also reported the first case of breast GCT in 1932 [2]. The 5th edition of the World Health Organization classification of breast tumors considers GCTs as mostly benign neoplasms with neuroectodermal origin, derived from Schwann cells [3]. Rare cases of granular cell malignant tumours are reported in the literature, associated with poor prognosis with lymph node and lung metastases [4].
Breast tumours represent 8% of all GCTs, which are more common in the head and neck, proximal extremities, gastrointestinal and respiratory tracts [3]. Usually, breast GCTs are single, but as many as 18% can be multicentric [3]. They arise more often in young adult African-American women, with earlier presentation in this population (mean age: 41|years as compared with 54|years in white-Americans) [3].
The majority of cases occur in the upper inner quadrant and it is suggested that GCTs grow through intracanalicular branches of the supraclavicular nerve, around Schwann cells [5].
Most cases are sporadic but breast GCTs have been reported in association with other conditions in Bannayan–Riley–Ruvalcaba syndrome, neurofibromatosis type|1 and Noonan syndrome (also known as LEOPARD syndrome) [3]. There are also reports of breast GCT associated with mastectomy scars and simultaneous GCT and breast carcinoma [3].
Loss-of-function mutations in ATP6AP1 and ATP6AP2 genes, involved in endosomal pH regulation, are frequently found in GCTs, leading to the accumulation of intracytoplasmic granules [3, 6]. These mutations are found irrespective of anatomic localization and are present in both benign and malignant tumours. As loss-of-function mutations in ATP6AP1 and ATP6AP2 genes were not yet described in other tumours, as far as it is known, they seem to be pathognomonic of GCTs [6].

Diagnosis

According to recommendations by EUSOMA (the European Society of Breast Cancer Specialists), in the preoperative phase, histopathology evaluation of core biopsy is the gold standard for GCT diagnosis [5], as clinical and radiological features of breast GCTs mimic breast carcinoma [3, 7].
In fact, GCTs present as irregular firm masses, typically superficial and mobile, but may rarely be adherent to the pectoralis fascia. They may cause skin thickening, retraction and nipple inversion [3]. On mammography, ultrasound and magnetic resonance imaging they are poorly defined masses with spiked margins but no microcalcifications [3, 7]. Despite these features usually associated with carcinomas only 1–2% of breast GCT cases show histological malignant change [3]. It is uncertain if malignant GCTs are malignant transformations from benign lesions or occur ‘de novo’ [4].

Pathologic findings

1. Macroscopic appearance

Grossly, GCTs are white or tan, firm and homogeneous, with regular or spiked margins, that can reach up to 5|cm (Fig.|1) [3].

2. Histopathologic features

Different sampling methods allow histological evaluation of GCTs: (1) needle-core biopsy (Fig.|2a); (2) vacuum-assisted breast biopsy (Fig.|2b); (3) surgical specimens, such as lumpectomies (Fig.|2c).
Histologically, breast GCTs are poorly defined tumours with infiltrative borders (Fig.|2c), composed of sheets, clusters or trabeculae of large, round and polygonal cells, separated by collagenous bands (Fig.|3a). The cells have indistinct borders and may have a syncytial pattern (Fig.|3b).
The hallmark feature is the presence of abundant eosinophilic cytoplasm with granular appearance (Fig.|3b). The nucleus is centrally located and it is usually round, small, hyperchromatic, rarely vesicular and with prominent nucleoli. Mitoses are usually absent. Perineural and perivascular invasion is frequently present. When GCTS are localized in the dermis, they may be associated with pseudoepitheliomatous hyperplasia. In small and superficial cutaneous biopsies, this sometimes can be confused with squamous cell carcinoma [3, 6].
The finely granular cytoplasm is derived from lysosome accumulation. Larger intracytoplasmic granules with clear haloes, named pustulo-ovoid bodies of Milian, are usually periodic acid–Schiff stain (PAS) positive and diastase resistant [3, 6].
The rare malignant GCTs morphologically vary from having a sarcomatous appearance to relatively bland features [6]. The histological criteria suggestive of malignancy are: increased cellularity, spindling, high nuclear to cytoplasm ratio, marked pleomorphism, vesicular nuclei with prominent nucleoli, more than two mitoses per 2|mm2 and geographical necrosis [4]. Larger tumours (>5|cm) and local recurrence are also features favouring malignancy [6].
Differential diagnosis
The histological differential diagnosis of breast GCTs includes reactive histiocytic lesions, dermatofibroma, epithelial tumours such as apocrine neoplasms and invasive carcinomas, melanocytic lesions (nevi and melanoma), hibernoma and alveolar soft part sarcoma [3, 6].
Morphological and immunohistochemical characterization are crucial in the differential diagnosis.
GCTs are strong and diffusely immunoreactive for S100 protein (Fig.|4a), CD68, CD63 antigen and neuron-specific enolase (NSE). The cells are also positive for transcription factor SOX-10 (SOX-10) (Fig.|4b), calretinin and inhibin A, and show diffuse nuclear expression of transcription factor E3 (TFE3) and microphthalmiaassociated transcription factor (MTIF). The cytoplasmatic granules are PAS positive and diastase resistant. Usually, there is no expression of cytokeratins (Fig.|4c), glial fibrillary acidic protein (GFAP), melanoma antigen recognized by T-cells 1 (Melan-A), HMB45- reactive antigen (HMB45), estrogen and progesterone receptors and receptor tyrosine-protein kinase erbB-2 (ERBB2) [3, 6].
S100 protein and SOX-10 are markers of Schwann cells but they may also be expressed in melanocytic lesions and primary breast carcinomas. The absence of cytokeratin expression excludes an epithelial lesion and the absence of Melan-A and HMB45 expression renders less likely a melanocytic lesion.
NSE is also a marker of neuroectodermal cells and is not specific for GCTs.
Inflammatory cells such as histiocytes can express CD68, as well as CD63 antigen; the last may also be present in neural derived tissue.
Dermatofibroma is also a benign infiltrative lesion commonly located at the dermis or subcutis with an identifiable grenz zone. It has spindle cells with scant cytoplasm and elongated nuclei dispersed among collagen bundles. Unlike GCTs, the cells are negative for S100 protein, SOX-10 and NSE.
Hibernoma is a benign, richly-vascularized adipose neoplasm composed by large brown fat cells with eosinophilic or pale multivacuolated cytoplasm that is granular with central nucleus, admixed with white adipose tissue. The cells are also positive for S100 protein, but staining is more variable. Unlike GCT, hibernoma is platelet endothelial cell adhesion molecule (CD31) positive.
Alveolar soft part sarcoma is a rare tumour of the soft tissue. It has an alveolar architecture and it is composed of large epithelioid polygonal cells with granular eosinophilic cytoplasm and prominent nucleoli. It has strong nuclear expression of TFE3, and may be focally positive for S100 protein; however, in contrast to GCTs, the cytoplasm has crystalline material and the cells are positive for actin and desmin. Furthermore, it is characterized by an ASPSCR1-TFE3 fusion gene.

3. Cytology

Fine-needle aspiration (FNA) is often inconclusive. Smears are hypercellular, with large polygonal cells with fragile membranes and abundant eosinophilic granular cytoplasm and small regular nuclei; no stromal or myoepithelial components are present (Fig.|5) [3, 8].
The cytoplasm features can frequently be interpreted as apocrine benign lesions, histiocytic lesions or even apocrine, lobular and secretory invasive carcinomas [3, 8]. In contrast to apocrine cells, the cells of GCTs are larger, with more granular cytoplasm and poorly defined borders. Some cases may also have prominent nuclear atypia. The use of cellblock and immunohistochemistry can be crucial.
Moreover, owing to the rarity of breast GCTs, pathologists lack experience in its cytological evaluation [8]. This method is no longer used for diagnosis of breast GCTs [5, 8].

Treatment

Treatment of GCTs relies on local excision [3, 5]. The best approach is lumpectomy (Fig.|2c) or even excision by vacuum-assisted breast biopsy in small lesions, a relatively safe and minimally invasive procedure (Fig.|2b) [3, 9]. Sentinel lymph node biopsy is not recommended but may be considered in the rare cases of malignant GCTs [5]. Metastasis have been described in 50% of malignant GCTs [6].
The recurrence rate is very low, even when excised with positive margins.
There is no need for adjuvant therapy in the benign cases, but long-term follow-up is recommended [5].

Summary

GCTs of the breast are rare neoplasms with neuroectodermal origin. Despite the fact that they are mostly benign tumours, they can mimic malignancies owing to their clinical and radiological features. Histopathology evaluation is the gold standard for diagnosis. Morphological and immunophenotypical features are characteristic, however differential diagnoses must be kept in mind.
Multidisciplinary approach is essential in breast tumours, and close contact between clinicians, radiologists and pathologists is vital for the correct management of patients.

The authors

Juliana Filipe* MD, Daniela Vinha Pereira MD, Saudade
André MD Serviço de Anatomia Patológica, Instituto Português
de Oncologia de Lisboa Francisco Gentil, Lisboa, Portugal

*Corresponding author
E-mail: jffilipe91@gmail.com