CCL22 mutations in large granular lymphocytic leukemia

Large granular lymphocytic (LGL) leukemia is an indolent, mature cytotoxic T-cell or NK-cell proliferative disorder that is often associated with immune-mediated complications, including cytopenia and autoimmune diseases.^1,2 The genetic landscape of LGL leukemia has been elucidated, including frequent somatic mutations in STAT3, STAT5B and TET2 genes.³-⁹ Recently, somatic mutations in C-C motif chemokine ligand 22 (CCL22) were identified in patients with NK-cell LGL leukemia (NK-LGLL),¹⁰ suggesting a novel mechanism of LGL proliferation in mutated chemokine-induced microenvironmental alterations. However, information on the mutational profile of CCL22 and its clinical significance in LGL leukemia remains limited, and its genetic-phenotypic associations remain to be clarified. We investigated mutational profiles, including CCL22 gene mutations, in patients with LGL leukemia and identified CCL22 mutations in T-cell LGL (T-LGL) leukemia and NK-LGLL, and found that their associations with clinical features varied among the subtypes of LGL leukemia.

The study cohort included 23 patients with NK-LGLL and 52 with T-LGL leukemia. Immunophenotypes of T-LGL leukemia included CD8⁺TCRαβ (N=36), TCRγδ (N=7), and CD4⁺TCRαβ (N=9). This study was conducted in line with the Declaration of Helsinki and written informed consent was obtained from all patients. The study protocol was approved by the institutional review board of each participating institution.

We identified CCL22 mutations in two of the 52 patients (3.9%) with T-LGL leukemia (Table 1; Figure 1A) using target sequencing (Ion Ampliseq Technology, Thermo Fischer Scientific), in addition to NK-LGL leukemia. Case 24 had CD8⁺ TCR αβ type-LGL leukemia with a STAT3 D661Y co-mutation of variant alle frequency (VAF) at 2.3%. Case 25 had CD4⁺TCR αβ type-LGL leukemia with STAT5B Y665F and S TAT 3 K658R co-mutations. Notably, they shared the same CCL22 mutation site, P46R, with VAF of 4.0% and 1.6%, respectively (Figure 1B). The CCL22 mutation at P46R was confirmed by allele-specific polymerase chain reaction (PCR) in both patients (representative data of case 24 are shown in Online Supplementary Figure S1). Sorted CD8⁺ T cells in case 24 were negative for CCL22 mutation but positive for S TAT 3 D661Y mutation. No cutaneous or neurological manifestations were observed in either of the patients.

Recurrent CCL22 mutations in six of the 23 patients (26%) with NK-LGLL were at L45R (N=3) and L45Q (N=3) (Figure 1; Online Supplementary Table S1). The median VAF of CCL22 mutations was 33.2% (range, 9.9-41.5%). Among the six patients with NK-LGLL, two presented with neurological symptoms and one presented with neutropenia. None of the six patients had anemia. TET2 mutations were identified in three of the 23 NK-LGLL cases, with a median VAF of 20.9%. All three patients with TET2 mutations had cytopenia and two had STAT 3 co-mutations. Among NK-LGLL cases, CCL22 mutations were mutually exclusive of STAT 3 and TET2 mutations. The immunophenotypes of NK-LGLL cells were CD16⁺CD56^bright in CCL22-mutated patients and CD16⁺CD56^-/dim in STAT3- and/or TET2-mutated patients (P=0.00072).

We performed real-time PCR (RT-PCR) to search for CCL22 mRNA in the peripheral blood of patients with CCL22 L45R mutation (case 3) or L45Q mutation (case 4). RT-PCR identified CCL22 mRNA in all the cell fractions. The mutated sequence was detected in DNA from peripheral blood mononuclear cells (Online Supplementary Figure S2A). The mutated sequence was also detected in mRNA from peripheral blood mononuclear cells by allele-specific PCR for L45R (Online Supplementary Figure S2B), whereas mutated mRNA was not detected by Sanger sequencing (Online Supplementary Figure S2A). The sorted NK cells were positive for DNA mutations.

Table 1.CCL22-mutated T-cell large granular lymphocytic leukemia.

Figure 1.Mutational landscapes of large granular lymphocytic leukemia. (A) Heatmap of recurrently mutated genes in natural killer (NK) cell granular lymphocytic Leukemia and T-cell large granular lymphocytic leukemia (T-LGLL). (B) Loliplot of C-C motif chemokine ligand 22 (CCL22) mutation in LGLL.

Two of the eight patients with CCL22 mutations required medical intervention for LGL leukemia and/or its complications; one had NK-LGLL complicated by subcutaneous panniculitis-like T-cell lymphoma at the time of diagnosis (case 2). Oral prednisolone therapy improved the cutaneous lesions, and a significant decrease was observed in NK cells in the peripheral blood. The other patient had CD8⁺TCRαβ LGL leukemia (case 24), the diagnosis of which was preceded by a history of idiopathic pure red cell aplasia (PRCA) that had been in complete remission with cyclosporine and prednisolone for >10 years. Her anemia eventually relapsed, and CD8⁺TCRαβ LGL leukemia was diagnosed.

In this study, CCL22 mutations were detected in patients with both T-LGL and NK-LGLL. The VAF in T-LGL leukemia cases tended to be lower than those in NK-LGLL cases (Table 1; Online Supplementary Table S1). In a previous report describing the results of an extensive in silico analysis of 2,837 cases of various T/NK neoplasms, including T-LGL leukemia, CCL22 mutations were not identified.¹⁰ The immunophenotypes of CCL22-mutated T-LGL leukemia seemed variable; one patient had the CD4⁺TCRαβ type, while the other had the CD8⁺TCRαβ type. Interestingly, CCL22 and STAT 3 were co-mutated in two cases of T-LGL leukemia in our cohort. In contrast, CCL22 and STAT 3 were mutually exclusive in patients with NK-LGLL, which is consistent with a previous study.¹⁰ Collectively, our results imply that CCL22 mutations may not be limited to NK-LGLL and may also be prevalent in other T/NK neoplasms with a unique set of molecular and clinical features, although the frequency of CCL22 mutations differed significantly between NK-LGLL and T-LGL leukemia (26% vs. 3.8%; P=0.007), and VAF were lower in T-LGL leukemia, suggesting a different impact of CCL22 mutations on T-LGL leukemia.

CCL22 mutations in NK-LGLL resulted in a high VAF, implying that mutated NK cells represent a major clone. In contrast, the low burden of CCL22 mutations in T-LGL leukemia implies subclones of T-LGL leukemic cells or small clonal cells in non-T-LGL cells, perhaps NK cells. The CCL22-non-mutated T-LGL leukemic cells in case 24 support the latter possibility, although we were unable to confirm the CCL22-mutated cell lineage in the patient due to insufficient material. The frequencies of the reported CCL22 mutational sites in LGL leukemia were quite low (<0.00001) in the databases, including Togovar and gnomAD. Mutational sites in LGL leukemia are unique, although various CCL22 mutations have been reported in solid tumors¹¹ (COSMIC, accessed February 2024).

CCL22 mRNA expression was weak in both wild-type and CCL22-mutated NK cells, which implies that a low level of mutated CCL22 may be sufficient to play a pathophysiological role in NK-LGLL with CCL22 mutations. It is also possible that mutant CCL22 expression is enhanced in specific environments (e.g., bone marrow).

The genetic and phenotypic characteristics of CCL22-mutant NK-LGLL cases include infrequent cytopenia, predominant immunophenotype of CD16⁺CD56^bright NK cells, and absence of STAT3 or TET2 mutations.¹⁰ However, CCL22-mutated T-LGL leukemia presented unique characteristics with lower VAF levels and co-mutational status in STAT. In T-LGL leukemia with CCL22 mutations, the mutated CCL22 - originating from either non-leukemic cells or LGL cells bearing these mutations - may, in conjunction with activating mutations in STAT within LGL cells, contribute to LGL proliferation and the emergence of associated complications. The prevalence and pathophysiological implications of CCL22 mutations in individuals with LGL leukemia, other disorders, and within healthy populations, necessitate further investigation.

Footnotes

• Received March 14, 2024
• Accepted May 17, 2024

Correspondence

*YM and TK contributed equally as first authors

F. ISHIDA - fumishi@shinshu-u.ac.jp

Disclosures

No conflicts of interest to disclose.

Funding

This work was supported in part by Kaken20K080709 and 21K16302 from a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan..

Acknowledgments

The authors would like to thank Dr. Keiki Nagaharu of Kuwana City Medical Center, Dr. Takeki Mitsui of Gunma University, Dr. Yukio Hirabayashi of Matsumoto Medical Center, Dr. Toshiharu Yujiri of Yamaguchi University, and Dr. Hiroyuki Takamatsu of Kanazawa University for providing the samples and clinical data.