Myeloid/Lymphoid Neoplasms Associated With Eosinophilia and Rearrangements of PDGFRA, PDGFRB, or FGFR1 or With PCM1-JAK2

Abstract

Objectives

To summarize cases submitted to the 2019 Society for Hematopathology/European Association for Haematopathology Workshop under the category of myeloid/lymphoid neoplasms with eosinophilia and PDGFRA, PDGFRB, or FGFR1 or with PCM1-JAK2 rearrangements, focusing on recent updates and relevant practice findings.

Methods

The cases were summarized according to their respective gene rearrangement to illustrate the spectrum of clinical, laboratory, and histopathology manifestations and to explore the appropriate molecular genetic tests.

Results

Disease presentations were heterogeneous, including myeloproliferative neoplasms (MPNs), myelodysplastic syndromes (MDSs), MDS/MPN, acute myeloid leukemia, acute B- or T-lymphoblastic lymphoma/acute lymphoblastic lymphoma (ALL/LBL), or mixed-lineage neoplasms. Frequent extramedullary involvement occurred. Eosinophilia was common but not invariably present. With the advancement of RNA sequencing, cryptic rearrangements were recognized in genes other than PDGFRA. Additional somatic mutations were more frequent in the FGFR1-rearranged cases. Cases with B-ALL presentations differed from Philadelphia-like B-ALL by the presence of an underlying MPN. Cases with FLT3 and ABL1 rearrangements could be potential candidates for future inclusion in this category.

Conclusions

Accurate diagnosis and classification of this category of myeloid/lymphoid neoplasms has important therapeutic implications. With the large number of submitted cases, we expand our understanding of these rare neoplasms and improve our ability to diagnose these genetically defined disorders.

Myeloid and lymphoid neoplasms with eosinophilia (M/LNs-Eo) and rearrangements of PDGFRA, PDGFRB, and FGFR1 were recognized as a stand-alone category in the 2008 World Health Organization (WHO) classification. PCM1-JAK2 and its genetic variants have subsequently been added to this category as a provisional entity in the 2016 WHO classification.^1,2 The diagnosis and classification of these entities require demonstration of the specific gene fusions, which lead to overexpression of an aberrant tyrosine kinase or cytokine receptor. The common presentation is a myeloproliferative neoplasm (MPN) with associated eosinophilia; however, the disease manifestations are heterogeneous and complex, potentially involving myeloid and/or lymphoid lineages, concomitantly or sequentially occurring in the same patient.

In this workshop, we received 53 cases submitted under this category, including 22 cases with PDGFRA rearrangements, 13 cases with PDGFRB rearrangement, 13 cases with PCM1-JAK2 rearrangements, and 5 cases with FGFR1 rearrangements. In addition, four cases with FLT3, ABL1, and LYN rearrangements, initially submitted to the “myeloid neoplasms associated with eosinophilia” session, are also discussed here. Many of these cases provided detailed clinical information, histopathologic images of bone marrow (BM), peripheral blood (PB) and extramedullary tissue, treatment and follow-up data, and the methods used for fusion gene detections. Next-generation sequencing (NGS) studies for somatic mutations were also available in a considerable number of cases. By summarizing this large number of cases, we will expand our understanding of these rare neoplasms and improve our capabilities to identify and diagnose these genetically defined neoplasms. Accurate diagnosis and classification have important therapeutic implications. In addition, these workshop cases raise good questions on when and which molecular genetic tests should be performed. These cases also illustrate overlaps and differences from Ph-like B-lymphoblastic lymphoma/acute lymphoblastic leukemia (B-LBL/ALL) harboring the same genetic abnormalities. This summary will incorporate the recent updates, discuss the caveats and pitfalls, and make appropriate recommendations for practicing pathologists.

Myeloid/Lymphoid Neoplasms With Eosinophilia Associated With PDGFRA Rearrangement

Rearrangement of PDGFRA is the most frequently found genetic abnormality within the group of M/LNs-Eo. The most common rearrangement is the FIP1L1-PDGFRA fusion due to a cryptic deletion at chromosome 4q12. Patients frequently present with eosinophilia, many with hypereosinophilia, defined as an absolute eosinophil count (AEC) of 1.5 × 10⁹/L or more; however, cases with only mild or absent peripheral blood eosinophilia have also been reported.³

Twenty-two cases with PDGFRA rearrangement were submitted to the 2019 Society for Hematopathology workshop under the category of M/LN-Eo, which constituted 42% of all cases submitted to this category Table 1 and Table 2. Two cases (case 157 and case 251) were reclassified by the panel. Case 157 was a 77-year old man who developed acute myeloid leukemia (AML) with eosinophilia 15 months posttreatment for B-lymphoblastic leukemia/lymphoma (B-ALL/LBL). FIP1L1-PDGFRA was detected within a complex karyotype in the AML BM sample but not in the B-ALL BM sample. It is uncertain if this represented a therapy-related AML or clonal evolution of B-ALL. Of note, AML with PDGFRA rearrangement as a subtype of M/LN-Eo typically shows a normal or noncomplex karyotype. Therefore, this case was not thought to be characteristic of a bona fide M/LN-Eo with PDGFRA rearrangement. Case 251 was a 26-year-old man with no significant medical history who was diagnosed with B-ALL with 5% peripheral blood eosinophils. Karyotype was abnormal, 46,XY,dup(1)(q21q32),der(16)t(11;16)(q13;p13.3), but without 4q12 abnormalities. NGS-fusion panel (RNA-seq) revealed RUNX1 and PHF6 mutations and incidental FIP1L1-PDGFRA rearrangement that was later confirmed by fluorescence in situ hybridization (FISH). The patient was treated with dasatinib and hyper-CVAD (cyclophosphamide, vincristine, daunorubicin, dexamethasone), achieved morphologic remission, but had persistent minimal residual disease. PDGFRA fusion signals by FISH corresponded to the degree of involvement by B-ALL, and there was no underlying chronic myeloid neoplasm. This B-ALL was apparently de novo and best classified as Ph-like B-ALL rather than M/LN-Eo.

Clinical Presentations Are Variable but Frequently Associated With Eosinophilia

Of the 20 patients with PDGFRA-rearranged M/LN-Eo, there were 19 men and only 1 woman (case 001). The age at presentation ranged from 27 to 82 years, with a median age of 51 years. While a strong male dominance was similar to previous reports, the median age appeared to be older.^4,5 The common clinical presentation included leukocytosis with a median WBC count of 13.9 × 10⁹/L (range, 4.0-106.0 × 10⁹/L) and eosinophilia with a median AEC of 4.69 × 10⁹/L (range, 0.35 to 81.0 × 10⁹/L). Hypereosinophilia with an AEC of 1.5 × 10⁹/L or more was present in 15 cases, mild eosinophilia in 2 (AEC of 0.9 and 1.0 × 10⁹/L) cases, and no eosinophilia in 3 (15%) cases.

If blinded to the genetic information, the clinical and histopathologic diagnoses of the 20 cases would be chronic eosinophilic leukemia (CEL) (n = 6), extramedullary myeloid tumor associated with concomitant BM myeloid neoplasms (CEL, MPN-U, or chronic myeloid neoplasms that were difficult to further categorize) (n = 6 and one with no BM information), systemic mastocytosis (SM) with hypereosinophilia (n = 2), AML (n = 1), extramedullary B-LBL/ALL (n = 1), and T-lymphoblastic lymphoma (T-LBL/ALL) (n = 3). One of the T-ALL cases also presented with a mature T-cell lymphoma (case 001). These cases reflect the spectrum of M/LN-Eo–PDGFRA reported in literature,^6-9 with CEL being the most common, followed by others resembling atypical chronic myeloid leukemia (CML), BCR-ABL1 negative, T- or B-LBL/ALL, and, rarely, AML or T-cell lymphoma.

Bone Marrow and Peripheral Blood Findings

Review of the peripheral blood smears from patients with eosinophilia (n = 17), including both mild eosinophilia cases, demonstrated abnormal eosinophil morphology in 13 (77%) cases. The abnormal morphology of eosinophils was characterized by cytoplasmic degranulation, abnormal granulation, cytoplasmic vacuolization, and/or abnormal nuclear segmentation or lobulation (Image 1; case 142).¹⁰ Monocytosis was not a prominent feature in PDGFRA-rearranged M/LN-Eo, and the monocyte percentage in peripheral blood (PB) ranged from 1% to 8%. Only one case (case 188) showed relative monocytosis of 18% (absolute monocyte count of 0.88 × 109/L) and would fit better with MPN-U if blinded to the cytogenetic information. Interestingly, this case lacked peripheral eosinophilia and demonstrated an abnormal karyotype with t(4;22)(q12;q11.2), resulting in a novel variant SPECC1L-PDGFRA rearrangement.

The typical BM findings included hypercellularity due to a myeloid proliferation with increased eosinophils. Eosinophils on the bone marrow aspirate smear ranged from 13% to 40% of total nucleated cells, were more pronounced in cases of CEL, and were less prominent in other types of myeloid neoplasms. Megakaryocytes ranged from increased to normal to decreased in numbers; morphologically, they could be normal, MPN-like, MDS-like, or mixed MDS/MPN-like. BM fibrosis was present in eight (53%) cases. One case (case 088) had extensive bone marrow necrosis but lacked Charcot-Leyden crystals. The only case of AML (case 107 BM) showed 30% blasts in a background of dysgranulopoiesis accompanied by mild peripheral blood eosinophilia (AEC of 1.0 × 10⁹/L). Karyotype showed 45-46,XY,t(4;12)(q12;p13)[20], and FISH confirmed the ETV6-PDGFRA rearrangement.

Tryptase and/or CD117 immunohistochemical stains were performed in 10 cases; atypical mast cells, distributed singly and in loose aggregates, with round to spindle-shaped morphology and variable CD25 coexpression, were seen in six (60%) cases (Image 2; case 225). In two cases (case 087 and case 0144), the presence of spindle-shaped, hypogranular, dense aggregates of atypical mast cells with aberrant CD25 coexpression (Image 2) rendered an initial diagnosis of SM. The presence of eosinophilia prompted further studies for the recurrent gene fusions, which identified PDGFRA rearrangement and led to a diagnosis of M/N-Eo with PDGFRA rearrangement. Both patients tested negative for KIT D816V and showed excellent response to tyrosine kinase inhibitor (TKI) therapy with imatinib. Similar cases defined as SM-CEL associated with FIP1L1-PDGFRA have consistently lacked the KIT D816V mutation.^11,12 Because of the absence of a mast cell driver mutation, the attribution of these rare cases to the WHO classification of systemic mastocytosis remains problematic. As such, it is highly recommended to test cases of SM for the presence of PDGFRA rearrangement when the KIT D816V mutation is absent.

Extramedullary Involvement/Presentation Is Common

Myeloid sarcoma/extramedullary involvement was thought to be uncommon in patients with PDGFRA-rearranged M/LN-Eo, with only a handful of cases reported to date prior to the workshop.¹³ However, 11 (52%) submitted workshop cases reported extramedullary lesions, suggesting this may be a more frequent phenomenon than previously recognized. The most common extramedullary site was lymph node (LN) (six cases) followed by epidural or spinal masses (four cases). Cutaneous involvement and an oral mass were seen in one case each. Of the latter, one case (case 001) demonstrated lesions in LN and skin. Infiltration with maturing myeloid elements with associated eosinophilia was the most common finding (six cases), often with various degrees of fibrosis, resulting in crush artifact Image 3A and Image 3B. Some cases demonstrated extensive necrosis Image 3C. Three cases were composed predominantly of immature cells, either myeloid sarcoma Image 3D, Image 3E, and Image 3F or monocytic sarcoma. T-LBL/ALL was found in three cases (case 001 also had a T-cell lymphoma) and B-LBL/ALL in one case. A concomitant BM myeloid neoplasm was seen in nine patients, including eight CEL cases and one MPN-U case. It is important to mention that eosinophilia was not a prominent feature in cases with T-LBL, B-LBL, or T-cell lymphoma, and the diagnosis of the M/LN-Eo with PDGFRA rearrangement was suspected either based on the BM findings or incidentally discovered by RNA sequencing analysis. These cases illustrate the importance of BM examination, the necessity of including M/LN-Eo in the differential diagnosis, and maintaining a low threshold for FISH or RNA-seq testing in the assessment of these extramedullary tumors.

PDFGRA Rearrangements Are Usually Cytogenetically Cryptic

The most common genetic variant in the group of M/LN-Eo and PDGFRA rearrangement is the FIP1L1-PDGFRA fusion,¹⁴ resulting from a submicroscopic 800-kb interstitial deletion on chromosome 4, del(4)(q12q12), in 17 (85%) of 20 cases (Table 2). The current FISH test for FIP1L1-PDGFRA is performed by using a three-probe set (SCFD2, LNX, PDGFRA) wherein a deletion of LNX (CHIC2) with retention of the flanking SCFD2 and PDGFRA is indicative of the FIP1L1-PDGFRA fusion.¹⁵ Of the 17 cases with FIP1L1-PDGFRA rearrangement, 16 cases were detected by FISH. One case (case 212) was negative by FISH, but a large CHIC2 deletion was detected by single-nucleotide polymorphism copy number microarray analysis. This case highlights the sensitivity and limitations of the FISH assay and indicates the need for alternative methods in cases with a high index of suspicion.

Seven additional fusion partners of PDGFRA have been described, including BCR, ETV6, KIF5B, CDK5RAP2, STRN, TNKS2, and FOXP1.¹ Rearrangements involving these variant partner genes are often not cryptic, exhibiting 4q12 abnormalities by conventional cytogenetics, which should be confirmed by FISH. Three workshop cases showed variant PDGFRA rearrangements: t(4;12)(q12;p13)/ETV6-PDGFRA (case 107), t(4;22)(q12;q11.2)/SPECC1L-PDGFRA (case 188), and one case with no karyotype information but a KIF5B-PDGFRA rearrangement by RNA-seq (case 142). While the cases with ETV6-PDGFRA and SPECC1L-PDGFRA had minimal peripheral blood eosinophilia, case 142 exhibited prominent PB and BM eosinophilia, as well as an erythroblastic sarcoma with numerous eosinophils in a lymph node.

Additional karyotypic abnormalities were detected in three cases: der(18)t(8q;18p), del(5q), and tetrasomy 8. NGS analysis was performed in four of the submitted cases, and two (50%) cases showed mutations (Table 2).

M/LN-Eo With PDGFRA Rearrangements Demonstrate Excellent Response to Imatinib

TKI imatinib is used as a first-line treatment, leading to durable hematologic and molecular response in most patients with M/LN-Eo with PDGFRA rearrangement. Complete hematologic remission is usually achieved in almost all patients within 1 month, with molecular remission after a median of 3 months of treatment. In patients who do not respond within 1 month of monotherapy, imatinib should be discontinued, and other therapies should be considered. The maintenance therapy is administered at a reduced dose to all responders, as discontinuation of imatinib can lead to relapse, and similar to chronic myeloid leukemia, ongoing treatment is recommended in patients with the FIP1L1-PDGFRA–positive neoplasms.¹⁶ Treatment with imatinib was reported in 13 workshop cases—monotherapy in 11 patients and in combination with ALL and AML therapy in 2 patients. Twelve (92%) patients achieved excellent clinical response. Of note, like FIP1L1-PDGFRA cases, those with variant rearrangements are also sensitive to imatinib therapy.¹⁴

Myeloid/Lymphoid Neoplasms With Eosinophilia Associated With PDGFRB Rearrangement

Thirteen PDGFRB-rearranged cases were received, constituting 25% of all cases submitted to the M/LN-Eo session (Tables 1 and 2). Case 283 was a 61-year-old woman with an initial diagnosis of MDS. Approximately 5 to 6 years later, she developed T-LBL/ALL in a lymph node that was successfully treated; however, 8 months later, she developed AML with a complex karyotype harboring an 11q23/KMT2A rearrangement. Although there were no 5q31-33 abnormalities or associated eosinophilia, PDGFRB rearrangement by FISH was performed, showing the presence of the rearrangement in the AML. Retrospective testing showed PDGFRB in the T-LBL/ALL but not in the MDS. This case likely represented clonal evolution and not a de novo M/LN-Eo with PDGFRB rearrangement. The remaining 12 cases included 11 men and 1 woman, with a median age of 51 years (range, 26-86 years). The demographic features were similar to the published literature that M/LN-Eo with PDGFRB rearrangements is considerably more common in men, with the peak incidence occurring in the fourth decade.⁵

Disease Presentations

Leukocytosis and hypereosinophilia were common laboratory findings with a median WBC of 34.4 × 10⁹/L (6.8-116 × 10⁹/L) and a median AEC of 4.44 × 10⁹/L (0.07-73.5 × 10⁹/L). Hypereosinophilia with an AEC of 1.5 × 10⁹/L or more was present in nine (75%) cases (Table 1). PB eosinophil morphology was reported in nine cases, and five (56%) of them demonstrated abnormal features, such as uneven granulation, cytoplasmic vacuolization, or abnormal nuclear segmentation or lobulation. Monocytosis has been described as a common finding in PDGFRB-rearranged cases, and a combination of monocytosis and eosinophilia (chronic myelomonocytic leukemia with eosinophilia [CMML-Eo]) is highly suggestive of the presence of PDGFRB rearrangement.¹⁷ Monocytes in the submitted 12 patients ranged from 1% to 21% in the PB with an absolute monocyte count ranging from 0.07 to 8.49 × 10⁹/L. Only two patients (case 205 and case 078) had relative (≥10%) as well as an absolute monocytosis (≥1.0 × 10⁹/L).

Disease presentations according to clinical and histopathologic features included CEL (n = 6), CMML-Eo (n = 2), SM associated with MDS (n = 1), MPN-U (n = 1), T-LBL/ALL in LN and CEL in BM (n = 1), and chronic basophilic leukemia (n = 1). The spectrum of disease presentations was similar to what has been reported of PDGFRB-rearranged cases,^17,18 which also include atypical chronic myeloid leukemia (aCML), MDS/MPN-U, and, rarely, AML. Notably, there were no cases of B-LBL/ALL submitted, likely due to a reclassification of such cases as Ph-like B-ALL.^19-21

Bone Marrow Morphology Is Uniformly Abnormal

The typical bone marrow morphology showed hypercellularity for age with a granulocytic proliferation and eosinophilia. Eosinophils in the BM aspirate smears ranged from 8% to 46% in cases where the differential count data were available. MDS-like megakaryocytes (small hypolobated and forms with abnormal nuclear lobation) were reported in seven (58%) cases. In the remaining five cases, the megakaryocyte morphology was described as unremarkable. No cases reported MPN-like megakaryocytes. Fibrosis, grade 2 (MF2) or higher, was seen in four cases. Mast cell workup was performed in six cases, and three (50%) cases showed spindle-shaped mast cells, scattered or in loose aggregates, with aberrant CD25 coexpression (case 052, case 078, and case 232).

PDGFRB Rearrangement Can Be Cryptic

The PDGFRB gene is located at 5q31~33, and more than 30 partner genes have been described to date.²² The most common genetic variant is t(5;12)(q32;p13.2)/ETV6-PDGFRB. PDGFRB rearrangements with alternative partners such as EBF1, SSBP2, TNIP1, ZEB2, and ATF7IP often present as de novo B-ALL, which is currently classified as Ph-like B-ALL^19-21 and not M/LN-Eo. On the other hand, in B-ALL with PDGFRB fusion to ETV6 or other partner genes, a differential diagnosis between M/LN-Eo and Ph-like B-ALL would be determined based on the presence or absence of an underlying myeloid neoplasm Figure 1.

The current dogma postulates that conventional cytogenetic analysis with sufficient metaphases should show 5q31-33 abnormalities in cases of PDGFRB-rearranged M/LN-Eo. Surprisingly, six (50%) submitted cases showed no 5q31~33 karyotypic abnormalities (five cases had a normal karyotype and one case showed del(12)(q24.1q24.3)). In one additional case (case 232) that presented as SM-MDS, inv(5)(q13q31) was missed in the first diagnostic BM, likely due to the subtle change of this structural abnormality. RNA-seq was performed in three cases and identified the partner genes as BCR (case 078), AFAP1L1 (case 052), and SART3 (case 162), respectively. One could argue that the high rate of cytogenetically cryptic PDGFRB-rearranged cases in the submission may reflect a submission bias; however, a recent study showed that among cases that underwent a systemic screen by PDGFRB FISH, two of four PDGFRB rearrangements were cryptic.²³ Furthermore, in case 052, a case of chronic basophilic leukemia with a normal karyotype, conventional FISH failed to detect the PDGFRB rearrangement, but RNA-seq revealed AFAP1L1-PDGFRB due to an intrinsic inversion. This case confirms the previous report that FISH is not 100% sensitive in demonstrating rearrangement of PDGFRB.²⁴ These cases echo the recent awareness of cryptic PDGFRB rearrangements in the era of increased RNA sequencing in clinical samples²⁵ and suggest the need to lower the threshold for PDGFRB FISH and/or RNA-seq testing, especially in cases with variant partner genes other than ETV6.

NGS analysis was performed in eight cases and reported myeloid-associated pathogenic mutations in four (50%) cases, with ASXL1 being the most common (three cases).

M/LN-Eo With PDGFRB Rearrangements Demonstrate Excellent Response to Imatinib

Similar to PDGRFA-rearranged cases, patients with M/LN-Eo with PDGFRB rearrangement demonstrate an excellent response to imatinib.¹⁶ Treatment with imatinib was reported in five cases, with all five patients achieving excellent clinical response. In case 162 and case 232, both patients received a trial of low-dose imatinib for refractory diseases and achieved excellent response. PDGFRB rearrangement was retrospectively performed and discovered by RNA-seq and FISH, respectively. In fact, low-dose imatinib has been used in patients with persistent eosinophilia but no specific molecular genetic lesions²⁶; complete responses may represent diagnostically occult PDGFRA or PDGFRB rearrangement or other unknown pathogenic targets.²⁷ Therefore, an empirical trial of imatinib may be considered in patients with hypereosinophilia not responsive to conventional therapy, and retrospective testing (RNA-seq) is recommended in responders.

Myeloid/Lymphoid Neoplasms With Eosinophilia Associated With FGFR1 Rearrangement

M/LN-Eo with FGFR1 rearrangement was initially described as 8p11 myeloproliferative syndrome due to the frequent involvement of the short arm of chromosome 8p11.^28,29 The disease is known for its complex presentations, including MPN, MDS/MPN, AML, T- or B-ALL/LBL, or mixed-phenotype acute leukemia, frequently associated with PB and/or BM eosinophilia. In the 2016 revised WHO classification, this disease is defined as (1) MPN or MDS/MPN with prominent eosinophilia and sometimes with neutrophilia or monocytosis or (2) AML, T- or B-ALL, or mixed-phenotype acute leukemia (usually associated with peripheral blood or bone marrow eosinophilia) and (3) the presence of t(8;13)(p11.2;q12) or a variant translocation leading to FGFR1 rearrangement, demonstrated in myeloid cells, lymphoblasts, or both.¹

We received five cases with FGFR1 rearrangements (Tables 1 and 2). Albeit a small number of cases, they highlighted the characteristics of this disease.

Clinical Presentation Is Very Complex

The FGFR1-rearranged cases published in the literature reported a wide range of age (3-84 years), with a median age of 30 to 40 years and a slight male predominance (1.5:1).^5,29 The demographic features of the five cases submitted to the workshop were similar to that published in literature, including three men and two women aged 31, 33, 51, 58, and 59 years.

Clinically, four patients had hypereosinophilia (≥1.5 × 10⁹/L); one patient had no eosinophilia but left-shifted granulocytosis (case 290). Four of five patients had extramedullary disease, including two cases of T-LBL (case 79 and case 290), one case of leukemia cutis (case 99), and one case of myeloid sarcoma (case 131). In the study by Strati et al³⁰ consisting of 17 patients from one institution, 35% of FGFR1-rearranged cases had concomitant LBL in lymph nodes and a chronic myeloid neoplasm in BM. Of note, some of these lymph nodes could have contained a minor myeloid component in addition to LBL, a feature that should further raise suspicion of a FGFR1 rearrangement.^31-33 PB and BM may show features reminiscent of CML, CEL, MPN-unclassifiable, aCML, or MDS/MPN. In case 290, a mast cell proliferation became evident in the post therapy BM, meeting the diagnostic criteria for SM. Similar to PDGFRA and PDGFRB cases, scattered spindle mast cells with aberrant CD25 expression are frequent, but only rare cases would meet the criteria for SM.^34-37 Such cases may be mistakenly diagnosed as SM with an associated hematologic neoplasm (SM-AHN) but should be classified as a FGFR1-rearranged neoplasm.

It is important to recognize that in some patients with an initial presentation of AML or ALL, an underlying chronic myeloid neoplasm may not be recognized at the initial diagnosis but only brought to attention when an 8p11 translocation persistently appears in a remission bone marrow following induction chemotherapy.^38,39

Karyotype Is Often Not Cryptic but Imprecise; Partner Genes May Contribute to Disease Phenotype

FGFR1 rearrangements are not cryptic; as expected, all five submitted cases exhibited 8p abnormalities (Table 2). However, the reported bands ranged from 8p11.2, 8p12 to 8p21, likely attributable to the relatively small size of band 8p11. Depending on the quality of chromosomal morphology, conventional G-banding may not precisely locate the involved subbands of the short arm of chromosome 8. On the other hand, of the myeloid and/or lymphoid neoplasms with 8p11 rearrangement, only about half of the cases harbor a FGFR1 rearrangement.³⁰ These caveats indicate the need of confirmatory testing, ideally for all myeloid or lymphoid neoplasms with rearrangement involving 8p11-21.

To date, there are 14 gene partners reported to fuse with FGFR1,^{28,36,38,40-50} and some of the partner genes may play a role in the different disease phenotypes. M/LN-Eo with t(8;13)/ZMYM2-FGFR1, the most common rearrangement, was once named blastic T/myeloid neoplasm with ZMYM2-FGFR1⁵¹ due to its common presentation as MPN associated with eosinophilia, either with concurrent T-LBL or a rapid evolution to blastic T/myeloid neoplasm. Two of the five submitted cases had t(8;13)/ZMYM2-FGFR1, one presented with T-LBL and another with myeloid sarcoma (case 131 and case 290). Case 49 was diagnosed with B-lymphoblastic crisis of MPN, a typical example of the reported association between t(8;22)(p11.2;q11.2)/BCR-FGFR1 and a B-ALL presentation.^38,39 Patients with t(6;8)(q27;p11.2)/FGFR1OP-FGFR1 were reported to present with polycythemia,⁵² but this feature was not seen in case 79 with this specific rearrangement.

Somatic Mutations Are Common and Frequently Involve RUNX1

NGS was performed in all five cases. Case 49 had a RUNX1 mutation, and case 290 showed STAG2 mutations. Unlike PDGFRA, PDGRFB, and PCM1-JAK2, somatic mutations are common in FGFR1-rearranged cases and frequently involve RUNX1.^30,53 Patients with RUNX1 mutations often present with acute leukemia.³⁰

Prognosis Is Poor; FGFR Inhibitor Is in Clinical Trial

Treatment information was available in only two patients; both received hematopoietic stem cell transplantation (HSCT). One patient relapsed (case 49) and received the FGFR inhibitor pemigatinib. Unlike M/LN-Eo associated with PDGFRA and PDGFRB, FGFR1-rearranged neoplasms do not respond to first-generation TKI (imatinib) therapy. Prognosis is poor, and aggressive chemotherapy and HSCT are considered the best curative option.²⁹ Clinical trials using third-generation TKIs or small molecules are ongoing with some hopeful results.^15,54 Pemigatinib, an FGFR inhibitor, has demonstrated clinical efficacy in patients with M/LN-Eo with FGFR1 rearrangement in clinical trials and individual patients.⁵⁵

Myeloid/Lymphoid Neoplasm With Eosinophilia Associated With PCM1-JAK2 Rearrangement

M/LN-Eo associated with PCM1-JAK2 rearrangement^1,56 and its genetic variants have been included as a new provisional entity in the 2016 WHO revision.¹ These are rare neoplasms, ^57,58 and the disease spectrum and histopathologic and molecular genetic features remain to be explored and expanded.

Thirteen cases were submitted to the workshop under this category (Tables 1 and 2). By chromosomal rearrangement/gene fusion, nine cases had t(8;9)(p22;p24)/PCM1-JAK2, one had BCR-JAK2, one had ETV6-JAK2, and two had unknown partner genes (not studied). By clinical presentation, 11 cases were de novo, whereas 2 were acquired in patients with prior myeloid neoplasms (case 81 and case 136). Case 136 was a 33-year-old woman with refractory AML with a normal karyotype and NPM1 and FLT3 ITD mutations. Five months later, the BM showed persistent AML, and karyotype showed t(8;9)(p22;p24)/PCM1-JAK2, while the NPM1 and FLT3 mutations remained positive. Case 81 described a 31-year-old female patient who developed therapy-related MDS (t-MDS) following cytotoxic therapy for Hodgkin lymphoma. The t-MDS had a normal karyotype but showed eosinophilia in BM. She was subsequently diagnosed with breast cancer and received chemotherapy and radiation therapy. Approximately 1 year later, a lymph node biopsy specimen revealed T-LBL/ALL, and BM showed a chronic myeloid neoplasm with the classic “triad” histopathologic features (see the description under the histopathology features). PCM1-JAK2 was demonstrated in both the T-ALL and BM myeloid components. Of note, the prior t-MDS did not have the same cytogenetic abnormality or histopathology features. Although an unrelated hematopoietic neoplasm remained possible, such cases would be more appropriate to be considered as disease progression/clonal evolution than the entity of M/LN-Eo with PCM1-JAK2.

With this large number of cases, we organized them in a way to address several commonly encountered questions.

Can We Assume the Presence of PCM1-JAK2 Fusion With t(8;9)(p22;p24)?

Of the nine cases with t(8;9)(p22;p24) and/or PCM1-JAK2, eight showed t(8;9)(p22;p24), while one had no available karyotype (case 197). PCM1-JAK2 fusions were demonstrated in six patients, two by FISH with break-apart probes (case 21 and case 136), one by reverse transcription polymerase chain reaction (RT-PCR) (case 58), and three by RNA sequencing (case 81, case 197, and case 255). Case 70, case 122, and case 208 had an abnormal karyotype, including t(8;9)(p22;p24), but in case 70, only a JAK2 FISH was performed while no additional testing was available for the two other cases. The question was raised if these three cases could be considered M/LN-Eo with t(8;9)(p22;p24)/PCM1-JAK2 when PCM1-JAK2 fusion was not studied.

It is noteworthy that this entity was initially described in 1990 in patients with “Ph-negative neutrophilic myelofibrosis” with t(8;9)(p22;p24).⁵⁹ In 2005, PCM1-JAK2 fusion was discovered as a result of t(8;9)(p22;p24).⁵⁶ Since then, a number of cases with t(8;9)(p22;p24) have been studied for PCM1-JAK2, either by RT-PCR or by FISH using break-apart probes, and all demonstrated the PCM1-JAK2 fusion.^56,58,60,61 It is generally accepted that the presence of t(8;9)(p22;p24) in the right clinical context is presumptive evidence of PCM1-JAK2. However, since the JAK2 FISH probe is widely available and can be performed on interphase cells, it is highly recommended as a confirmatory study.

Genetic Variants of PCM1-JAK2

In the 2016 WHO classification, it is stated that cases with translocations resulting in a fusion gene between JAK2 and an alternative partner, specifically, t(9;12)(p24.1;p13.2)/ETV6-JAK2 and t(9;22)(p24.1;q11.2)/BCR-JAK2, may be considered as variants of this provisional entity. These cases are extremely rare, and information is limited.⁵⁷

Case 234 (BCR-JAK2) was a 39-year-old male patient presenting with a high WBC count with mild eosinophilia and basophilia, with blood features reminiscent of CML, chronic phase.⁶² Karyotype showed 46,XY,add(22)(q11.2)[13]/46,XY[7], with no 9q24 abnormality. BCR-JAK2 was detected by RNA sequencing. So far, fewer than 20 cases of BCR-JAK2 have been reported. Patients have a median age of 51 years (range, 2.7-67 years), a male predominance (2:1), and a common presentation of MPN, MDS/MPN, and, rarely, ALL or AML.^1,58,62-64 Case 13 was a 41-year-old man with a presentation that would otherwise be classified as CEL, NOS.⁶⁵ Karyotype showed der(9)t(9;12)(?p24;p13), and RNA sequencing confirmed the ETV6-JAK2 fusion. Prior to this case, only 11 cases of myeloid and/or lymphoid neoplasms with ETV6-JAK2 have been reported in the literature, including 8 cases of ALL (2 T-ALLs, 6 B-ALLs), 2 of MDS, and 1 of atypical CML.^57,66 The cases with a presentation of B-ALL may belong to Ph-like B-ALL,^21,67,68 leaving the real M/LN-Eo with ETV6-JAK2 extremely rare.

The partner genes were unknown in two cases, one with no karyotype available (case 247) and another with an uninformative karyotype reporting –Y only (case 260). Since only JAK2 FISH was performed, it was unclear if these two cases were PCM1-JAK2 or variants. Both cases had eosinophilia, with clinicopathologic features of CEL. Case 260 showed typical BM histopathologic “triad” features (see the description under the histopathology features). In the Tang et al⁵⁸ multicenter study of 13 JAK2-rearranged myeloid and/or lymphoid neoplasms, 4 (30%) cases showed a partner gene other than PCM1, one being BCR-JAK2, and the other three were classified as Ph-like B-ALL. It is reasonable to suggest that variants other than BCR-JAK2 and ETV6-JAK2 are extremely rare in M/LN-Eo. B-ALL with JAK2 rearranged to other partner genes, such as t(5;9)(q14;p24.1)/SSBP2-JAK2,^58,69 most likely belongs to Ph-like B-ALL.

Can JAK2 Rearrangement Be Cryptic and Can One Assume a JAK2 Rearrangement With t(9p24.1;v)?

In case 260, in which the karyotype showed –Y only, FISH with the JAK2/D9Z1 probe did not show abnormalities, but the JAK2-XL probe revealed an extra fusion signal, as well as two fusion signals of JAK2 in 79% of the cells, suggesting the presence of a JAK2 rearrangement and/or cryptic (segmental) aneusomy of chromosome 9. JAK2 is located at the terminal band of chromosome 9p. Since t(9p24.1;v) rearrangements involve only a very small fragment of 9p, the translocation can be cryptic, especially if the partner gene is also located at the terminal end of a chromosomal arm. This can be extremely problematic if the JAK2 rearrangement is due to an insertion. Thus, it is likely that JAK2 rearrangements may be underrecognized if the detection is based only on t(9p24.1;v). As RNA sequencing has been increasingly used in clinical practice, more cases with JAK2 rearrangements may be identified, especially those with variant partner genes.

On the other hand, a karyotype of t(9p24.1;v) other than t(8;9)(p22;p24), including add(9)p24, cannot be assumed as the presence of a JAK2 rearrangement. It has been shown that only a quarter of cases with t(9p24.1;v) actually harbor a JAK2 rearrangement,⁵⁸ indicating the need for FISH confirmation in these cases.

Disease Spectrum and BM Histopathologic Features of PCM1-JAK2

Of the cases with confirmed t(8;9)(p22;p24)/PCM1-JAK2, seven cases were detected at the time of diagnosis and two acquired at disease progression (case 81 and case 136). Of the former seven patients, one was a 1-year old infant who had de novo T-ALL with no underlying chronic myeloid neoplasm pre- or posttherapy (case 197); therefore, the case would be more appropriately classified as “T-lymphoblastic leukemia/lymphoma with PCM1-JAK2 fusion,” following a similar approach in the differentiation of Ph-like B-LL vs M/LN-Eo with a B-LL presentation. On the other hand, case 21 was a 69-year-old man with no history of a hematologic neoplasm who had “de novo” B-ALL; however, a background chronic myeloid neoplasm became evident after the patient achieved B-ALL remission. The most appropriate diagnosis and classification would be M/LN-Eo with t(8;9)(p22;p24)/PCM1-JAK2 presenting in B-lymphoblastic crisis of a chronic myeloid neoplasm. Of the remaining five patients, four had a chronic myeloid neoplasm in chronic phase (case 58, case 122, case 208, and case 255) and one in accelerated phase (case 70).

While eosinophilia was not present in the cases of T-ALL and B-ALL, eosinophilia was common in patients with chronic myeloid neoplasms (four of five patients), which was marked in three and mild in one. Organomegaly and/or lymphadenopathy were very common and present in all patients.

Histopathologically, the BM showed the typical “triad” Image 4, including hypercellularity with an eosinophilic infiltrate, large aggregates of immature erythroid precursors, and myelofibrosis. Similar features could also be seen in the extramedullary sites, containing large aggregates of immature erythroid precursors/pronormoblasts,⁵⁸ which represent extramedullary involvement by M/LN-Eo with PCM1-JAK2. In some cases, the immature erythroid precursors form large sheets, consistent with a diagnosis of myeloid sarcoma, erythroblastic type (case 255). Other findings, such as erythroid hyperplasia and prominent dyserythropoiesis, may be present. Some cases may show dysgranulopoiesis and dysmegakaryopoiesis.

The BM “triad” histopathologic features were not observed in cases with BCR-JAK2 and ETV6-JAK2 variants, a finding similar to cases published previously.^1,58,62-64 In contrast, case 247 and case 260, with unknown partner genes, had marked eosinophilia and showed strikingly similar histopathologic features to the cases with confirmed PCM1-JAK2 rearrangement, suggesting that the partner gene in these two cases might have been PCM1.

The Clinical Course and Prognosis Are Highly Variable

The disease courses were highly variable. For patients with a chronic phase myeloid neoplasm, the disease may be indolent, and patients may have a long life span,^57,58 such as case 208 and case 247. Two patients achieved normalization of the CBC with no treatment (case 260) or with only hydroxyurea cytoreduction (case 13). However, long-term follow-up in these cases was not available. On the other hand, some patients may progress to accelerated phase or acute leukemia and have a poor prognosis, such as case 58, case 70, and case 255. While response to imatinib is poor, targeted therapy with JAK2 inhibitors such as ruxolitinib may offer potential benefit.⁷⁰ Some patients have achieved excellent outcome after allogeneic stem cell transplant (case 122).^1,5

Somatic Mutations Are Less Common

NGS was performed in 11 of 13 cases. Not considering the case of persistent AML with NPM1 and FLT3 mutations (case 136) and the infant with de novo T-ALL, somatic mutations were detected in three (38%) of eight cases, including ASXL1, TET2, BCORL, and EP300. This finding is similar to the PDGFRA- and PDGFRB-rearranged cases, in which somatic mutations are reported less common than FGFR1-rearranged cases.^53,58

Myeloid/Lymphoid Neoplasms With Eosinophilia and Other Gene Rearrangements (Possible New Provisional Entities)

The workshop received four cases with previously described recurrent rearrangements involving FLT3, ABL1, and LYN, which could be possible candidates for future inclusion in the group of M/LN-Eo.

Most cases have CEL, often in association with T-LBL; thus, they seem to fit well with other types of M/LN-Eo. Fusion genes involving FLT3, such as ETV6-FLT3, have been reported in several studies.⁷¹ Cases of FLT3 fusion partners other than ETV6 have also been rarely reported. Case 150, genetically characterized by a t(13;14)(q12;q32)/TRIP11-FLT3, had T-LBL in a lymph node associated with a CEL picture in bone marrow and peripheral blood. In addition, the bone marrow showed 5% involvement by T-LBL. A similar case has been described recently.⁷² Other rare cases have been associated with SPTBN1-FLT3 and GOLGB1-FLT3 fusions.^73,74 A limited role for FLT3 inhibitors, mostly as a bridge to stem cell transplant (SCT), has been described. FMS-like tyrosine kinase 3 inhibitor (sunitinib) treatment has also been used in a few patients.⁷⁵ Case 116 and case 201 were cases with ETV6-ABL1 fusion. This fusion gene results from a t(9;12)(q34;p13) or complex rearrangement.⁷⁶ Because the creation of the ETV6-ABL1 fusion requires at least three chromosomal breaks, typically a cryptic insertion to generate an in-frame fusion gene, routine karyotype is usually inconclusive, and FISH can miss the small insertions. Only a combination of ETV6 and ABL1 probes, targeted RT-PCR, or RNA-seq can reliably identify the fusion.⁷⁷ Adult cases can present as chronic myeloid neoplasms or AML; all reported adult cases displayed eosinophilia. Both cases submitted to the workshop had eosinophilia: case 116 as chronic myeloid neoplasm and case 201 in transformation to AML. Pediatric cases, more commonly presenting as ALL, often lack an eosinophilic component. The prognosis in M/LN-Eo–ETV6-ABL1 is poor in adults with AML but better in patients with a CEL presentation. Dasatinib response was reported in case 116. Long-term remission (7 years) was described in a similar case by Xie et al.⁷⁸ Last, case 223 was an example of T-LBL involving lymph nodes and CEL in bone marrow and peripheral blood associated with the ETV6-LYN fusion. The case was cytogenetically silent, and the fusion was detected by RT-PCR (multiplexed PCR assay). A similar case has been reported in association with a t(8;12)(q12.1;p13) and a complex chromosomal rearrangement. This case, similar to case 223, required RT-PCR for a fusion partner identification.⁷⁹

Summary

Considering the rarity of myeloid/lymphoid neoplasms associated with eosinophilia and rearrangements of PDGFRA, PDGFRB, or FGFR1 or with PCM1-JAK2, the number of cases we received in this workshop is truly remarkable. These cases not only showed us the classic clinical, laboratory, and morphologic features of these neoplasms but also greatly expanded our understanding of the spectrum of these diseases. Remarkably, these cases were collected recently, at a time when advanced molecular genetic testing is becoming more routinely available in clinical practice. A fair amount of information is new, providing great insight into the pathogenesis, as well as demonstrating tools for better detection of these genetically defined neoplasms. The following key points summarize the essential information of this session.

1. M/LN-Eo and rearrangements of PDGFRA, PDGFRB, or FGFR1 or with PCM1-JAK2 is a historical term and also an “umbrella” category. A variable proportion of cases do not have, or only have, mild eosinophilia at presentation. Within each specific gene subcategory, the presentations are often highly variable, which may be partly due to different partner genes involved, additional molecular genetic hits, and require different management actions. Therefore, following the “umbrella” category, it is advisable to spell out the actual disease and partner genes (if known), as exemplified by the following terminology: myeloid/lymphoid neoplasm with eosinophilia and PDGFRA rearrangement (FIP1L1-PDGFRA) presenting as chronic eosinophilic leukemia, with extramedullary involvement or myeloid/lymphoid neoplasm with eosinophilia and FGFR1 rearrangement (BCR-FGFR1) presenting in B-lymphoblastic phase of chronic myeloid neoplasm.

2. PDGFRA, PDGFRB, or FGFR1 or with PCM1-JAK2 may acquire in the disease course of a well-established hematopoietic neoplasm, as disease evolution/progression, and should not be considered the entity of M/LN-Eo.

3. Disease presentations are heterogeneous and often complex. Histopathologically, the bone marrow is often abnormal, but the changes are variable from case to case except for cases with PCM1-JAK2 that frequently exhibit classic “triad” features. The BM diagnoses blinded to genetic information could be CEL, MPN-U, MDS/MPN (CMML, aCML, MDS/MPN-U), MDS, AML, B-ALL/LBL, T-ALL/LBL, or even SM, highlighting the importance of incorporating genetic results into the diagnosis as required when using the WHO classification. Extramedullary involvement is very common. The extramedullary lesions can range from very mature to very immature (myeloid sarcoma or lymphoblastic lymphoma), often accompanied by an eosinophilic infiltrate.

4. In addition to FIP1L1-PDGFRA, other rearrangements may be cryptic, especially in cases with PDGFRB rearranged to partner genes other than ETV6. On the other hand, karyotyping is neither precise nor synonymous with the presence of fusions with exception of t(8;9)(p22;p24), which is usually indicative of the presence of PCM1-JAK2. Nonetheless, FISH (or RNA-seq, RT-PCR) for PDGFRA, PDGFRB, FGFR1, and JAK2 is required for confirmation. RNA sequencing, either targeted or high-throughput NGS-based assay, has detected fusions in unsuspected cases, indicating the need for lowering the threshold to perform FISH and/or applying RNA-seq as a broader screening tool.

5. In the absence of widely available RNA-seq, FISH studies may be initiated on the following cases: (a) cases with eosinophilia that are neither reactive nor associated with genetically defined entities, such as CML or AML with inv(16); (b) cases with a complex clinical manifestation involving myeloid and lymphoid lineages or with extramedullary involvement with or without infiltrating eosinophils; (c) systemic mastocytosis with concomitant hypereosinophilia, or SM-AHN, when KIT mutations are absent; (d) cases with karyotypic abnormalities involving 4q12, 5q31-33, 9p24, or 8p11, regardless of eosinophilia; and (e) retrospective testing on all responders to empirical treatment with a trial of imatinib.

6. These entities overlap with Ph-like B-lymphoblastic leukemia with rearrangement involving PDGFRA, PDGFRB, FGFR1, and JAK2. A road map (Figure 1) is provided to show the basic principle in distinguishing these two categories of diseases. The key elements include the presence of a chronic myeloid neoplasm that manifests prior to, concomitant with, or after therapy for B-ALL, as well as involved partner genes.

7. There are other candidates for the family of M/LN-Eo, including fusions involving FLT3, ABL1, and LYN. More studies are needed for further characterization.

References