Epidemiology of malignant lymphoma in Japan

Incidence and mortality rates of malignant lymphoma in Japan

The Center for Cancer Control and Information Services of the National Cancer Center reported estimated malignant lymphoma incidences of 22.3/100,000 and 18.3/100,000 in men and women, respectively, in 2013, with mortality rates of 11.4100,000 and 8.5/100,000 [1]. Worldwide, the incidences of non-Hodgkin lymphoma were 6.0/100,000 and 4.1/100,000 in men and women, with mortalities of 3.2/100,000 and 2.0/100,000, respectively [2]. The number of patients with malignant lymphoma is reportedly increasing worldwide, including Japan, although the rate of increase differs in each country [3].

Proportions of subtypes in Japan

Although B-cell lymphomas tend to account for a large proportion of cases worldwide, the proportion varies depending on the region. The breakdown of subtypes of cases (N = 1442) diagnosed in Kurume University in Japan in 2014 is shown in Table 1. B-cell neoplasms, T/NK cell neoplasms, and Hodgkin lymphoma were observed in 70.8, 16.4, and 6.4% of the cases, respectively. Diffuse large B-cell lymphoma (DLBCL) occurred in 35.8% of cases, followed by follicular lymphoma (FL, 21.7%), adult T-cell leukemia/lymphoma (ATLL, 5.8%), angioimmunoblastic T-lymphoma (AITL, 4.8%), extranodal marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue (MALT lymphoma, 3.4%), mantle cell lymphoma (MCL, 3.3%), peripheral T-cell lymphoma, not otherwise specified (PTCL-NOS, 2.8%), and B-cell lymphoma, unclassifiable, with features intermediate between DLBCL and Burkitt lymphoma (1.7%).

Table 1 Frequency of subtypes of lymphoid neoplasms stratified by six areas in Japan
Full size table

Classification of data from six regions in Japan

To investigate the differences among topographic areas in Japan, Table 2 shows data classified into the Okinawa, Kyushu, Kinki/Chugoku/Shikoku, Chu-bu, Kantou, and Tohoku/Hokkaido regions. Although the proportion of ATLL was higher in Okinawa (16.8%) and Kyushu (9.2%) than those in the other areas, consistent with previous studies [4, 5], the proportions of the other subtypes did not differ significantly between regions.

Table 2 Comparison among studies of frequency of lymphoid neoplasm in Japan
Full size table

Comparison to previous reports in Japan

Table 2 shows comparisons with previous studies in Japan. The Lymphoma Study Group (LSG) of Japanese Pathologists (N = 3194; 1994-1996) [5], Chihara et al. (N = 26,141; 1993–2006) [6], and Aoki et al. (N = 2260; 2001–2006) [4] investigated Japanese data. Although there were few differences among subtypes, the proportions of FL in Aoki et al. (18.3%) and our study (21.7%) were higher than those in the LSG of Japanese Pathologists (6.7%) and Chihara et al. (5.1%). It is possible that temporal changes might have contributed to the higher proportion of FL. Additional studies with larger populations are necessary to investigate this trend because there were also differences in study design among these studies.

Comparison of the proportions of subtypes among China, Korea, and Japan

Comparison of the proportion of subtypes in East Asia, including China [7], Korea [8], and Japan revealed a higher proportion of FL and ATLL cases in Japan and a higher proportion of extranodal NK/T-cell lymphoma, nasal type cases in China (Table 3). Although the results of ATLL and extranodal NK/T-cell lymphoma are compatible to those with previous studies, to our knowledge, the difference in the proportion of FL cases has not been previously reported. The westernization of eating habits may be one of the reasons for this observation.

Table 3 Subtypes of lymphoid neoplasms in China, Korea, and Japan
Full size table

Recent research on adult T-cell leukemia/lymphoma in Japan

Large retrospective studies of ATLL

Katsuya et al. performed a large retrospective study of 1594 cases of ATLL from 84 institutions, including 895 acute type, 355 lymphoma type, 187 chronic type, and 157 smoldering type according to the Simoyama classification [9]. The median survival times (MSTs) and four-year overall survival (OS) were 8.3 months [95% confidence interval (CI) 7.5–8.9] and 11% for acute type, 10.6 months (95% CI 9.3–11.9) and 11% for lymphoma type, 31.5 months (95% CI 25.9–41.1) and 16% for chronic type, and 55.0 months (95% CI 36.6–90.4) and 52% for smoldering type, respectively.

Previous studies reported that allogeneic hematopoietic stem cell transplantation (allo-HSCT) improved prognosis in cases of acute and lymphoma types [10]. Katsuya et al. also observed that cases receiving allo-HSCT had an increased four-year OS of 26%. However, they also reported the significant effect of disease status at the time of allo-HSCT. Although the MSTs from transplantation are 22 months in cases with a first remission, in cases with primary refractory and relapse, the MSTs are only 3 months.

For favorable chronic and smoldering types, the MSTs and four-year OS were not reached (95% CI, 40.7 to not estimable) and 60% and 55.0 months (95% CI 36.6–90.4) and 52%, respectively. The MSTs and four-year OS for the unfavorable chronic type were 27.0 months (95% CI 20.4–35.0) and 29%. The median times from diagnosis to the first-line systemic chemotherapy were also shorter in the unfavorable chronic type (3.7 months, 95% CI 1.4–7.3) than in favorable chronic type (39.1 months (95% CI, 21.8 to not estimable) and smoldering type (56.0 months, 95% CI, 26.1 to not estimable).

Prognostic significance of soluble interleukin-2R in indolent ATLL

Katsuya et al. also investigated the prognostic factors of indolent ATLL in a retrospective study including 149 cases of chronic type and 118 cases of smoldering type ATLL [11]. Soluble interleukin-2 receptor (sIL-2R) was shown to be an independent prognostic factor among clinical features, including Shimoyama’s classification [12]. Based on the sIL-2R values, cases of indolent ATLL could be classified in low (sIL-2R < 1000 U/mL), intermediate (1000- < sIL-2R < 6000 U/mL), and high risk (sIL-2R- > 6000 U/mL). The MSTs and 4-year OS were not reached (95% CI 4.6-) and were 77.6% and 5.5 years (95% CI 3.1-), 54.1%, and 1.4 years (95% CI, 0.7–2.6) and 22.1% for low, intermediate, and high-risk, respectively (P < 0.0001).

This prognostic classification by sIL-2R can be also adapted as a prediction model for disease progression because the median times to systemic chemotherapy were 8.4, 2.7, and 0.1 years, respectively (P < 0.0001). Moreover, this model can statistically stratify cases of indolent ATLL for OS and median times to systemic chemotherapy more appropriately than Shimoyama’s classification based on favorable chronic, unfavorable chronic, and smoldering types. Thus, SIL-2R is expected to be used as a prognostic factor of indolent ATLL.

Genomic alterations associated with prognosis in aggressive and indolent ATLL

Based on the research from Katsuya et al. assessing clinical feature as a prognostic factor, Kataoka et al. evaluated 463 cases of ATLL to investigate the associations between genomic alterations and ATLL prognosis [13]. In 97% of cases, one or more somatic alterations, including those on phospholipase C, gamma 1 (PLCG1); protein kinase C beta (PRKCB); C–C chemokine receptor type 4 (CCR4); caspase recruitment domain-containing protein 11 (CARD11); signal transducer and activator of transcription 3 (STAT3), VAV1; tumor protein P53 (TP53); and transducin beta like 1 X-linked receptor 1 (TBL1XR1), were identified. The genetic profiles differed between aggressive type and indolent type. In comparison to the indolent type, the aggressive type showed an increased association with higher numbers of mutations, focal amplifications and deletions, hyperploid status, and cytosine–guanine dinucleotide island hypermethylation. Mutations in seven genes including PRKCB, TP53, interferon regulatory factor 4 (IRF4), interferon regulatory factor 2 binding protein 2 (IRF2BP2), tet methylcytosine dioxygenase 2 (TET2), CD58, and beta-2-microglobulin (B2M) and focal copy number alterations (CNAs) were more commonly detected in aggressive type than in indolent type. Among the genomic alterations, IRF4 and STAT3 mutations were the most significant in aggressive and indolent types, respectively.

Kataoka et al. then proposed a prognostic model that included both genetic abnormality and clinical features. In aggressive type ATLL, PRKCB mutations and PD-L1 amplifications were independent prognostic factors for poor OS as well as Japan Clinical Oncology Group Prognostic Index (JCOG-PI) high-risk categorization and older age (70 years or more). Cases of aggressive type were categorized into three groups according to the number of factors comprising age (70 years or more), PRKCB mutations, and PD-L1 amplifications. There was a significant difference in one-year OS rates among the three groups, with rates of 58% for cases with no risk factors, 45% for those with one risk factor, and 16% for those with two or more risk factors (P < 0.001). In indolent type ATLL, IRF4 mutations, PD-L1 amplifications, and CDKN2A deletions were independent prognostic factors for poor OS and the unfavorable chronic subtype in Shimoyama’s classification was not significant. Cases of indolent type were categorized into two groups depending on the presence or absence of at least one factor among IRF4 mutations, PD-L1 amplifications, and CDKN2A deletions. There was a significant difference in three-year OS rates between the group including cases with no risk factors (82%) and those with at least one risk factor (30%) (P < 0.001).

This prognostic model is outstanding because it considers both clinical features and genetic abnormalities of the aggressive and indolent types. This model might be adopted in clinical practice for ATLL.

Biological significance of RHOA mutations in ATLL

RHOA mutations are commonly observed in angioimmunoblastic T-cell lymphoma (AITL) [14]. Nagata et al. investigated RHOA mutations in 203 cases of ATLL [15]. Although AITL cases with RHOA mutations always have TET2 mutations [14], TET2 mutations were less often detected (17%) in RHOA-mutated cases of ATLL. The RHOA mutations in ATLL showed various patterns of distribution in the GTP-binding pocket. Although the primary mutation in AITL is Gly17Val [14], the mutation hotspots in ATLL include those at the Cys16, Gly17, and Ala161 residues, among which Cys16Arg mutations were most frequent. Although the Gly17Val mutant in AITL showed little binding to GTP [14], mutants in ATLL bound to GTP more immediately than did wild-type (WT) RHOA. The disconnection of GTP and GDP was highly promoted in the ATLL mutants. These results suggest that RHOA mutations in ATLL could increase the GDP/GTP exchange rate.

Flow cytometric analyses of ATLL cells with various RHOA mutations were performed to investigate whether differences in RHOA mutations affected the cell of origin of neoplastic cells. ATLL cells with WT RHOA, Cys16Arg, Cys16Gly, and Ala161Pro mutations were likely to be regulatory T-cells or effector T-cells due to their CD4(+), CD25(+), FoxP3(+), PD-1(−) or CD4(+), CD25(+), FOXP3(−), PD-1(−) phenotypes. On the other hand, ATLL cells with Gly17Val were considered to be memory T-cells because of their CD4(+) CD25(-) phenotypes.

Therapeutic effects of lenalidomide in ATLL

Although systemic intensive chemotherapy has been administered for the treatment of ATLL [16], satisfactory effects have not been achieved, especially in aggressive ATLL. Ishida et al. performed a multicenter Phase II study in which lenalidomide was administered for relapsed or recurrent cases of ATLL [17]. An overall response rate of 42% was achieved, including 15% of cases with complete remission (CR), 4% unconfirmed complete remission (CRu), and 23% partial remission (PR). The ORRs were 33% cases of acute type, 57% in lymphoma type, and 50% in unfavorable chronic type. The median time to response, median time to progression, mean duration of response, median progression-free survival, and median OS were 1.9, 3.8, 5.2, 3.8, and 20.3 months, respectively. As described above, the administration of lenalidomide showed a favorable therapeutic effect for relapsed or recurrent cases of ATLL. Lenalidomide might be a possible future option for the treatment of ATLL.

Novel molecular agents in ATLL

Although various therapeutic approaches, including conventional chemotherapy, molecular-targeted agents, anti-viral drugs, and allo-HSCT have gradually improved the prognosis of ATLL, new developments are also underway. Narita et al. reported that BAY1143572, a selective inhibitor of cyclin-dependent kinase 9 (CDK9), could be an effective drug for ATLL following examination of cell lines, ATLL cells from patients, and a mouse model [18]. BAY1143572 suppressed the proliferation in a concentration-dependent manner and induced apoptosis of ATLL-derived or human T-lymphotropic virus type 1 (HTLV-1)-transformed cell lines and ATLL cells from patients by inhibiting the phosphorylation of the Ser2 site in RNAPII. The expressions of the c-Myc and Mcl-1 proteins were also down-regulated. BAY1143572 did not affect the protein expression of HTLV-1 tax and induced various changes in protein expression of HTLV-1 basic leucine zipper (HBZ). The administration of BAY1143572 after injection of ATLL cells in mice caused the suppression of tumor development, decreased liver and bone marrow tumors, reduced sIL-2R levels, and significantly prolonged OS.

Immune checkpoints in ATLL

The biological and clinical significance of tumor immunity has recently been confirmed in hematological malignancy [19] as well as in solid tumor [20, 21]. The importance of programmed cell death ligand 1 (PD-L1) has been reported in ATLL [22]. The reported cases of ATLL included those with PD-L1 expression in neoplastic cells (nPD-L1 + ATLL), those with PD-L1 expression in stromal cells (miPD-L1 + ATLL), and those without PD-L1 expression (PD-L1 − ATLL) (Fig. 1). nPD-L1 + ATLL (MST = 7.5 months) had inferior OS compared to that of nPD-L1 − ATLL (MST = 14.5 months) (P = 0.0085). Among nPD-L1 − ATLL, miPD-L1 + ATLL (MST = 18.6 months) showed superior OS compared with that of PD-L1 − ATLL (MST = 10.2 months) (P = 0.0029). The expression of nPD-L1 and miPD-L1 maintained its prognostic value for OS in multivariate analysis (P = 0.0322 and P = 0.0014, respectively). In nPD-L1 + ATLL, the PD-1/PD-L1 pathway may contribute to a worse prognosis as well as other malignancies. Blockade therapy of the PD-1/PD-L1 pathway might improve the prognosis of nPD-L1 + ATLL. However, in miPD-L1 + ATLL, viral infection including HTLV-1 can induce PD-L1 expression in stromal cells. The reason why miPD-L1 + ATLL shows a better prognosis is unknown; however, the PD-1/PD-L1 pathway might also function in miPD-L1 + ATLL. If the effector function works more in miPD-L1 + ATLL, blockade therapy may result in much better prognosis. The membranous expression of HLA and β2M in ATLL is also associated with better prognosis and might reflect the immune response [23]. The immune checkpoints could be closely associated with ATLL pathogenesis and progression.

Fig. 1
figure 1

Criteria according to PD-L1 expression in ATLL. If PD-L1 was expressed on neoplastic cells, the case was categorized as nPD-L1-positive ATLL. If PD-L1 was not expressed on neoplastic cells, the case was considered nPD-L1-negative ATLL. Among nPD-L1-negative ATLL, if stromal cells showed PD-L1 expression, the case was considered miPD-L1-positive ATLL. If the stromal cells did not express PD-L1, the case was considered PD-L1-negative ATLL

Full size image