CRFL2 rearrangements may associate with underlying leukemic plasticity, raising concern for an inherent predilection to lineage switch following B-ALL-directed targeted therapies.

Despite tremendous successes of CD19-targeted immunotherapy in relapsed/refractory (r/r) B-lymphoblastic leukemia (B-ALL), lineage switch (LS) to acute myeloid leukemia (AML) is increasingly recognized as a form of immune escape.1–3 Broadly defined as immunophenotypic shift from one leukemia lineage to another with retention of baseline genetics, LS, and the mechanisms influencing its emergence, may vary based on presenting disease subtype. The propensity of infant B-ALL with KMT2A rearrangements (KMT2Ar) to undergo LS following B-ALL directed chemotherapy is well-established; however, the incidence of KMT2Ar-independent LS is increasing following targeted immunotherapy.1 4

Cytokine receptor-like factor 2 rearrangements (CRLF2r) are well-defined drivers of cell proliferation in B-ALL, resulting in constitutive JAK2 signaling with a Philadelphia (Ph)-like gene expression profile and chemotherapy resistance.5 However, alterations in CRLF2 are rare in AML and, to the best of our knowledge, have not been implicated in previously published cases of LS,6 thus the incidence of LS in the context of CRLF2 is unknown.

We present two cases of CRLF2r B-ALL that transitioned to AML following CD19-targeting. We discuss the important role of next-generation sequencing (NGS) and how extramedullary disease (EMD) may contribute to immune escape.

A patient in their 20s with r/r B-ALL with an IgH::CRLF2 fusion was referred for and consented to enrollment in a phase 1 study of CD19/CD22 chimeric antigen receptor (CAR) T-cells (NCT03448393) for refractory disease. Notably, the patient had EMD at diagnosis and at CAR T-cell infusion (table 1). Prior therapies include standard chemotherapy, along with blinatumomab and inotuzumab ozogamicin.

Case-based clinical presentation, diagnosis, and management of lineage switch

Pre-infusion, multiparameter flow cytometry (MFC) identified 27% of mononuclear cells in the bone marrow (BM) to be B-ALL, and F-18 FDG-PET/CT revealed concern for EMD due to splenomegaly and retroperitoneal lymphadenopathy with significant avidity (figure 1). The postinfusion course was complicated by grade 2 cytokine release syndrome, and grade 4 neutropenia and thrombocytopenia. Due to profound cytopenias, an early BM assessment was performed on day +21. This revealed minimal residual disease (MRD) negative complete remission in the BM by MFC, with incomplete count recovery and a markedly hypocellular marrow. Restaging FDG-PET on day +30 showed near resolution of prior EMD, however, an asymptomatic FDG-avid pancreaticoduodenal lesion (SUV 15.5) was identified (figure 1). The patient subsequently developed severe abdominal pain, direct hyperbilirubinemia, and hepatic transaminitis indicative of biliary obstruction, necessitating endoscopic retrograde cholangiopancreatography for stent placement. Severe thrombocytopenia precluded biopsy. The patient was empirically started on dexamethasone for treatment of presumptive EMD (B-ALL).

Cases 1 and 2 flow, NGS, pathology, and imaging. Case 1: (A) Disease burden by multiparametric flow cytometry and next-generation sequencing (NGS) by ClonoSEQ during the peri-CAR period. (B. Leukemic blasts on BM aspirate. At day −1 blasts are of variable size, contain scant basophilic cytoplasm, and fine chromatin representative of B-lymphoblasts. At day +48 monoblasts are uniformly large in size and contain abundant, prominently vacuolated cytoplasm and loose chromatin. (C and D) H&E and CD79a stains of BM biopsy specimens from day −1, day +21, and day +48. At day −1 there is an interstitial infiltrate of CD79a positive B-cells consistent with B-lymphoblastic leukemia. At day +21 the marrow is markedly hypocellular with no evidence of increased blasts and CD79a positive cells are essentially absent. At day +48 there are small foci containing an atypical mononuclear infiltrate consistent with blasts which are negative for CD79a, but positive or CD68 (not shown). (E) F-18 FDG-PET/CT imaging on day −2 demonstrating extensive EMD including peri-nephric and retroperitoneal node involvement, day +30 showing full resolution of kidney disease and retroperitoneal disease, but with emergence of new pancreatic mass, and day +62 demonstrating marked interval disease progression with extensive abnormalities involving the liver, pancreas, upper abdominal and retroperitoneal lymph nodes, internal mammary nodes, and axial and appendicular BM. Case 2: (A) Disease burden by multiparametric flow cytometry and NGS by ClonoSEQ. (B) Leukemic blasts identified in BM aspirates. At diagnosis, blasts were intermediate in size with high nuclear-to-cytoplasmic ratios, fine chromatin, inconspicuous nucleoli, and scant deeply basophilic agranular cytoplasm, consistent with ALL. At lineage switch, blasts were large in size with irregular/folded nuclear contours, multiple small nucleoli, and moderate amounts of basophilic cytoplasm. (C and D) H&E and representative immunohistochemical stains. At diagnosis, BM showed a diffuse infiltrate of intermediate-sized blasts with smooth chromatin and small nucleoli; these blasts expressed CD79a, consistent with B-lymphoblastic leukemia. At lineage switch, skin biopsy demonstrated a diffuse dermal proliferation of large cells with irregular nuclear contours, open chromatin, conspicuous nucleoli, and finely granular amphophilic cytoplasm. The cells expressed CD163 consistent with monocytic differentiation. (E) Coronal and axial contrast enhanced CT and fused FDG PET/CT images demonstrate hypoenhancing renal lesions and mesenteric lymphadenopathy at the time of diagnosis; and hypoenhancing hepatic and splenic lesions and a destructive rib lesion at the time of lineage switch, all of which demonstrate significant radiotracer uptake. Coronal fused PET/CT image also demonstrates a hypermetabolic left femur lesion which is not well appreciated on CT. ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; BM, bone marrow; LOD, limit of detection; PB, peripheral blood.

Serial NGS testing of IgH/Igκ/Igλ/TCRG/TCRB genes using the ClonoSEQ platform was performed peri-CAR T-cell infusion for enhanced MRD monitoring. By day +21, BM NGS demonstrated near complete eradication of all but one likely background level TCRG sequence (0.0012%) and one Igκ sequence below the level of detection. Given concern for progressive EMD, serial peripheral blood (PB) NGS testing was implemented and demonstrated re-emergence of all baseline sequences (figure 1) to thresholds typically detectable by MFC (0.4%), yet concurrent PB MFC remained negative for B-ALL. BM assessment was repeated on day +48 with expanded MFC analysis to evaluate for myeloid markers revealing LS to AML with monoblastic differentiation with loss of CD19 and CD79a. Although the IgH::CRLF2 fusion and TSLPR were not detected in this or subsequent samples, retention of the JAK2 mutation and NGS confirmed clonal relationship to pre-CAR disease.

AML-directed treatment, combined with ruxolitinib, was initiated. Despite multiple approaches, the patient died of disease progression approximately 1 month following diagnosis of LS.

A patient in their 20s with high-risk B-ALL and P2RY8::CRLF2 fusion was initially enrolled in the Children’s Oncology Group trial AALL1732 (NCT03959085) for newly diagnosed B-ALL; the patient achieved an MRD negative remission at end of induction. Following consolidation, the BM remained MRD negative by MFC; however, NGS testing by ClonoSEQ demonstrated detectable disease at 0.029% by IgH/Igκ (figure 1).

Given high-risk features and detectable sequences by NGS at end of consolidation, the patient was taken off protocol; blinatumomab was initiated with plans to consolidate using allogeneic stem cell transplantation (allo-SCT). Mid-cycle marrow evaluation demonstrated continued MFC MRD negativity, but rising MRD (0.537%) by NGS prompted expanded immunophenotyping. Post-blinatumomab BM using MFC incorporating myeloid markers identified an aberrant CD33+ monoblastic population enumerated at 0.38% of total cells consistent with LS to AML. Concurrent NGS showed a higher disease level of 4.5% prompting repeat marrow evaluation, including NGS of flow-sorted white blood cell populations, to clarify the MFC and NGS MRD discordance. MFC identified 1.9% atypical monocytes (of total non-erythroid cells), while the IgH/Igκ sequences were detected in 12.4% of total cells and specifically 97% of the monocytic/monoblastic population, confirming the presence of the clonal monocytic population. Notably, BM biopsy at this time showed 20% atypical monocytes by morphology suggesting sampling may have contributed to differences in residual disease.

Following LS, the patient demonstrated evidence of EMD involving skin, soft tissue, cortical bone, liver, and spleen by CT and FDG-PET (figure 1). Biopsies of skin and rib lesions were positive for CD4, CD68, CD163, and CD33 consistent with myeloid sarcoma with monocytic differentiation. Although the P2RY8::CRLF2 fusion was not detected by RNA fusionplex testing (despite robust CRLF2 expression by flow cytometry at diagnosis suggesting that the vast majority of blasts likely harbored the fusion) from the BM (~10% blasts) or rib biopsy, IgH/Igκ clonal sequences were retained, supporting the diagnosis of LS. Despite multiple rounds of myeloid-directed therapy, the patient experienced disease progression and died from AML 6.5 months after LS diagnosis.

Discussion

With the rapidly expanding landscape of antigen-targeted therapies, LS is increasingly recognized as a means of immune escape with dismal outcomes.1 3 Early suspicion for and detection of LS is critical for therapeutic management. Herein, we present two cases of LS arising in patients with CRLF2r following CD19-targeted immunotherapies. To our knowledge, these are the first cases of LS in CRLF2r B-ALL reported.

While we could not determine whether CRLF2r itself predisposed to lineage plasticity seen in these cases3 or if it was a passenger mutation without a causal relationship, these cases serve to raise awareness of the potential of LS developing in such patients. Further study is needed to determine if CRLF2r should be added to the growing list of non-KMT2Ar cytogenetic aberrations associated with LS (eg, TCF3::ZNF384) and the relationship thereof.2 6 7 Indeed, reports of myeloid propensity of CRLF2r B-ALL at diagnosis, and/or in relapse have also been described, although rarely.8–11 Specifically, myeloperoxidase expression meeting criteria for a diagnosis of mixed phenotype acute leukemia (MPAL, B/Myeloid) has been reported in a P2RY8::CRLF2 positive leukemia with a concurrent JAK1 mutation (referenced as “Ph-like MPAL,”) as has the myeloid-associated antigen CD66c8 9.

While retention of IgH/Igκ/TCRG/TCRB sequences at LS in these cases confirms transformation of the original lymphoid clone, the absence of CRLF2r at LS diagnosis in both cases is perplexing. This may represent inadequate detection of the fusions, particularly in case 1 in which the JAK2 mutation was retained. However, it raises the possibility that CRLF2r was not the LS/relapse-driving lesion, and that the other cytogenetic and/or molecular aberrations (such as the JAK2/PTPN11 in case 1 or the activating KRAS and WT1 mutation in case 2) may be more contributory. In fact, studies have demonstrated that P2RY8::CRLF2 fusions can occur as either early or late events in leukemogenesis, are frequently present in only small clones at initial diagnosis, and are commonly lost at relapse.12 13

These cases also illustrate the value of multi-modal disease surveillance. In both, MRD negativity by B-ALL-targeted MFC belied active disease progression. In the United States, MFC is most used for MRD determination in B-ALL. However, in LS, antigenic changes within a clonal population can preclude MFC detection if evaluation is limited to the original immunophenotype. Expanded MFC with myeloid antigen assessment should be considered following antigen-targeted therapy.

In both cases, NGS testing by ClonoSEQ demonstrated a discrepant rise in baseline sequences compared with MFC that foreshadowed LS. NGS utilization for MRD determination is evolving and currently being employed to guide risk stratification in upfront treatment protocols (NCT03914625). Further, NGS is increasingly used to monitor r/r B-ALL, particularly after CAR T-cell therapy and/or prior to allo-SCT.14 An upcoming prospective study will explore the use of NGS testing following CAR T-cells to guide management in pediatric B-ALL (NCT05621291).

Although serial imaging is not standard in B-ALL, in both cases, specialized imaging revealed significant EMD. Non-CNS EMD is increasingly recognized among children and young adults with multiply r/r B-ALL, with as high as 21.1% having isolated non-CNS EMD or combined medullary/non-CNS EMD detected on FDG PET-CT prior to CAR infusion.15 The incidence of EMD following antigen-targeted therapy is unknown, but may be similar to that seen following allo-SCT where less potent graft-versus-leukemia surveillance in extramedullary sites may contribute to EMD relapse.16 Accordingly, a high index of suspicion for EMD and low threshold for incorporation of serial imaging may be important additions to disease surveillance, particularly in monitoring for LS. A study at the National Cancer Institute (NCT05969002) aims to evaluate the incidence of non-CNS EMD in patients with r/r B-ALL using FDG PET-CT pre- and post-CAR T-cell infusion.

Ultimately, these cases illustrate the need for ongoing study of novel drivers of LS and close monitoring of patients with r/r B-ALL who receive immunotherapy. Future efforts to establish post-immunotherapy MFC monitoring using myeloid markers, incorporation of NGS, and serial imaging to evaluate for EMD are needed to optimize outcomes.

Not applicable.

This study involves human participants. Ethics approval is not applicable; individual case reports without identifiable information do no require IRB approval at our respective institutions. Participants gave informed consent to participate in the study before taking part.

We gratefully acknowledge the participants, their families, and the providing care teams.