Epidemiology and secular trends of malignant lymphoma in Japan: Analysis of 9426 cases according to the World Health Organization classification

Abstract This study provides an overview of the epidemiology and secular trends of malignant lymphoma in Japan. Using data from clinics and hospitals throughout Japan, we analyzed 9426 cases of malignant lymphoma diagnosed in 2007‐2014. We show that the proportion of follicular lymphoma and methotrexate‐associated lymphoproliferative disorder increased during this time, as did the onset age for follicular lymphoma and diffuse large B‐cell lymphoma. Significant increases in onset age for follicular lymphoma and diffuse large B‐cell lymphoma were observed in both men and women (all P values <0.0001 except for P = 0.0448 for the latter disease in women). Further studies are required to determine the reasons for the higher proportion of and onset age for these lymphomas. Additionally, we believe that continued observation of these trends is necessary.


| INTRODUCTION
The epidemiological characteristics of malignant lymphoma (eg, proportion, sex ratio, and age of onset) differ along sociological, environmental, and racial lines. [1][2][3][4] In adult B-cell lymphoma, diffuse large B-cell lymphoma (DLBCL, 37%) and follicular lymphoma (FL, 29%) are the most common subtypes worldwide, followed by extranodal marginal zone lymphoma of mucosa-associated lymphoid tissue (MALT, 9%), mantle cell lymphoma (MCL, 7%), and chronic lymphoid leukemia/small lymphoid leukemia (CLL/ SLL, 12%). 5 However, these proportions vary from country to country. CLL/SLL, for example, is less common in Asia than in other parts of the world. 5 In adult T-cell lymphoma, peripheral T-cell lymphoma not otherwise specified (PTCL-NOS) and angioimmunoblastic T-cell lymphoma (AITL) are the most common subtypes worldwide. 6 According to the international peripheral T-cell and natural killer (NK)/T-cell lymphoma study, their relative frequencies are 25.9% and 18.5%, respectively. 6 Other common T-cell lymphoma subtypes include extranodal NK/T-cell lymphoma (10.4%), adult T-cell leukemia/lymphoma (ATLL, 9.6%), anaplastic lymphoma kinase-positive anaplastic large cell lymphoma (ALCL-ALK ＋ , 9.6%), and ALK-negative  (ALCL-ALK − , 5.5%). 6 These proportions vary according to geographic locations; extranodal NK/T-cell lymphoma and ATLL, for example, are more common in Asia than in Europe and the USA. 5 Epidemiological studies of malignant lymphoma in Japan have been performed by the Lymphoma Study Group of Japanese Pathologists (3194 cases in 2000) 7 and Aoki et al (2260 cases in 2008). 8 The latter study examined geographic distribution, age of onset, and sex ratio in cases diagnosed at Kurume University from 2001 to 2006 using data from clinics and hospitals throughout Japan. However, no paper on the secular trend of epidemiological factors in Japan has been published to date.
Herein, we present an overview of the epidemiology and secular trends of malignant lymphomas diagnosed from 2007 to 2014 in Japan.

| Tissue samples
We reviewed 19 332 clinically suspected cases of hematopoietic/lymphoid tissue malignancies that were diagnosed in our laboratory between 2007 and 2014. For diagnosis, biopsy or resection specimens were sent to us from 396 hospitals and clinics across Japan's 47 prefectures, ranging from the northernmost prefecture of Hokkaido to Okinawa, the southernmost prefecture. Clinical data were obtained from clinicians at the time of the initial diagnosis.
Histological confirmation was obtained in 11 428 of the 19 332 cases, with 9426 remaining after exclusion of cases with relapses or recurrences. Malignant lymphomas were diagnosed and analyzed in accordance with the 2008 World Health Organization (WHO) guidelines. We used three procedures (described below) for diagnosis: flow cytometry, immunohistochemistry, and DNA and RNA analysis.
All final diagnoses were made by one pathologist (K. O.), who is an authority on the pathology of lymphoid neoplasms. This study was approved by the Research Ethics Committee of Kurume University.

| Immunohistochemistry
For immunohistochemistry, formalin-fixed paraffin-embedded tissue and/or frozen sections were used. All tissues and sections were stained with H&E and immunostained for pan-B-cell (CD20) and pan-T-cell (CD45RO) markers. Ancillary immunostaining was performed when needed. Antibodies (in parentheses) to the following proteins or cells were used for immunohistochemistry:

| DNA and RNA analysis
The procedures for in situ hybridization for EBV-encoded RNA, 9 southern blotting, 10 and fluorescence in situ hybridization 11 were described in detail previously. Fluorescence in situ hybridization was conducted to identify genetic abnormalities strongly associated with specific malignant lymphomas.

| Histologic subtypes
The histological subtypes of the malignant lymphomas in our study are listed in Table 1, and the distribution of each subtype among the nine geographic areas in Japan is presented in Table 1 and Figures 1 and 2 Overall, DLBCL is the most common and FL is the second in frequency followed by adult T-cell leukemia/lymphoma (ATLL), AITL, MALT and PTCL, NOS. Among B-cell lymphoma, DLBCL was the most common (50.3%), followed by FL (31.2%), MALT (5.2%), and MCL (3.9%). This order also applied to the different regions, excepting Chubu and Tohoku/Hokkaido. In these two regions, MCL was more frequent than MALT lymphoma (Chubu: MALT, 2.9% and MCL, 5.1%; Tohoku/Hokkaido: MALT, 3.2% and MCL, 4.3%) (Figure 3).

| Age distribution
Most of the patients in this study were >15 years old because patients ≤15 years old are almost exclusively diagnosed at the National Center for Child Health and Development in Japan. Accordingly, some types of lymphomas that occur in children may be elder than real age. Median ages at diagnosis were 68 years for B-cell neoplasms, 70 years for T/NK-cell neoplasms, 60 years for Hodgkin lymphoma, 69 years for histiocytic/dendritic cell neoplasms, 71 years for composite lymphoma, and 69 years for immunodeficiency-associated lymphoproliferative disorder. Hence, with the exception of Hodgkin lymphoma, most major malignant lymphoma subtypes are diagnosed when patients are in their late 60s to 70s. The median and mean ages for the sub-major subtypes were as follows: precursor B-lymphoblastic leukemia/lymphoma, 31 and 32.06 years (n = 16); mediastinal B-cell lymphoma, 40 and 49.16 years (n = 44); unclassifiable B-cell lymphoma with features intermediate between those of DLBCL and classical Hodgkin lymphoma, 52 and 51.00 years (n = 6); precursor T-lymphoblastic

| DISCUSSION
Although most of the findings of this study confirmed those of previous reports, 7,8 some novel observations were made, as detailed below.

| FL
The proportion of FL is higher in Western countries (32% in Omaha, NE, USA; 31% in Vancouver, Canada; and 28% in London, UK) than in Asian regions (8% in Hong Kong, China). 2 In Japan, it has gradually increased: 6% in 1996-2000, 12 18.3% in 2000-2006,8% and 22.4% in 2007-2014 (the present study) (Figure 7). Hence, the proportion of FL in Japan is approaching that in Europe and the United States.
The onset age for FL in Asian countries is lower than that reported in international studies (59 years), 13 with 54 years in southwest China, 14 52 years in eastern India, 15 and 52.5 years in Korea. 16 Our study shows a gradual increase in the median and mean age of patients with FL in Japan (P < 0.0001). These findings show that the onset age for FL is shifting from an Asian pattern to a Western pattern for reasons unknown.

| DLBCL
Our study shows that onset age for DLBCL has increased in Japan (P = 0.0002) ( Figure 6B). The median onset age for DLBCL in Japan was 72 years (mean age, 69.56 years), which is much higher than the median ages in Korea (59 years), 17 China (55 years), 14 and internationally (64 years). 13 The reason for the increasing onset age for DLBCL in Japan is unclear, but one hypothesis is worth noting. The proportion of secondary DLBCL, which tends to develop at a more advanced age than primary DLBCL, may be increasing. Due to a universal care system, elderly patients in Japan often receive chemotherapy, including molecular targeting therapy. Consequently, those with low-grade malignant lymphomas survive longer, allowing more time for secondary DLBCL to arise.  In addition, although detailed risk factors of DLBCL have not yet been revealed, potential risk factors affecting the onset age for DLBCL include chronic inflammation 18,19 and an immunocompromised state due to advanced age, 20 acquired immunodeficiency syndrome, 21 primary immune disorders, 22 certain medications, 23 and transplants. 24

| ATLL
Adult T-cell leukemia/lymphoma is caused by HTLV-1, which is most prevalent in the southern regions of Japan, such as Kyushu and Okinawa. 25 Hence, the geographical distribution of ATLL differs throughout Japan.
The onset age for ATLL differed significantly according to the year of diagnosis (2007-2009, 2010-2012, or 2013-2014) (P = 0.002). Significant differences were also observed in analyses stratified for men (P = 0.0004), but not for women (P = 0.3149). There were no constant trends for both men and women ( Figure 6C); in men, the onset age for ATLL was lowest in 2007-2009, followed in order by 2013-2014 and 2010-2012. In women, onset age was lowest in 2010-2012, followed in order by 2013-2014 and 2007-2009. Thus, although the differences between the time periods were statistically significant, we considered them to be essentially nonsignificant.

| MTX-LPD
Methotrexate-associated lymphoproliferative disorder is an "other iatrogenic immunodeficiency-associated lymphoproliferative disorder" that occurs in patients treated with MTX. 5 In this study, the median age at which MTX-LPD was diagnosed was 69 years, and the M/F ratio was 0.49. This female predilection presumably reflects the frequency of autoimmune diseases in women, including rheumatoid arthritis (RA). 26 The number of MTX-LPD cases has increased since 2008 (Figure 8). Two events may account for this increase: first, its inclusion in the 4th edition of the WHO guidelines in 2008, which widened its recognition by clinicians and second, the increased use of MTX. MTX was indicated for rheumatoid arthritis in 1999 and become the drug of first alternative in 2011 in Japan.

| Strengths and limitations
The strengths of our study are as follows. First, to the best of our knowledge, it is the largest epidemiological analysis performed in Japan, with >9000 cases. Second, all cases were diagnosed by the same pathologist, thus eliminating the need for resolution of diagnostic differences. Third, because Japan consists of essentially one race, consideration of racial factors was unnecessary.
Our study did, however, have limitations. First, most of the specimens were submitted to us by hematologists. Because some malignant lymphomas (eg, MALT lymphomas) are also treated by gastroenterologists, ophthalmologists, and dermatologists, their frequencies may be greater than reported here. Second, the frequency of subtypes usually diagnosed via analysis of peripheral blood (eg, hairy cell leukemia, CLL/SLL, and acute types of ATLL) may have been underestimated because blood samples were not submitted to us. Third, because ATLL has several histological forms, measurement of HTLV-1 antibody levels and detection of monoclonal integration of the HTLV-1 provirus are necessary for an accurate diagnosis. 27 However, these assays were probably not performed in regions in which HTLV-1 is not endemic, potentially resulting in underestimation of ATLL frequency. Fourth, as noted above, patients ≤15 years old in Japan are almost exclusively diagnosed in a hospital specializing in pediatrics. Therefore, the true epidemiology of some lymphoma subtypes may not be reflected in our study. Fifth, although we received data for this study from throughout Japan, the proportion of data received from Kyushu and Okinawa was relatively high. Thus, there is a possible bias caused by geographical distribution. Table 2 shows a comparison between the present study and related epidemiological studies in Japan. Although the present study included many specimens from Kyushu and Okinawa, which reportedly have high numbers of ATLL cases, Table 2 shows that the proportion of T-cell lymphomas was relative low  compared to that reported in previous studies. The reason for this is unclear, but it may be due to an increased proportion of FL and DLBCL. Finally, knowledge of a patient's history of MTX therapy is essential for diagnosis of MTX-LPD. However, the accuracy of such information may vary from hospital to hospital. In Japan, MTX is often prescribed by orthopedic specialists and rheumatologists. Large hospitals have many departments (eg, Orthopedics and Rheumatology) that share information about MTX administration, and thus, this information is likely reliable. On the other hand, in clinics or small hospitals, information about MTX use may not be readily available because MTX may have been prescribed elsewhere. Thus, the frequency of MTX-LPD in small hospitals and clinics may be underestimated.

| CONCLUSION
This study shows that the proportion of FL in Japan has increased between 2007 and 2014, as has the onset age for FL and DLBCL. Future monitoring of this trend should continue in the future, and the reasons for these increases should be determined.