Hematologic malignancies are classified on the basis of the morphology, immunophenotype, cytogenetics, and clinical characteristics. The most recent classification suggested by the World Health Organization (WHO) have grouped myeloid malignancies into five categories: acute myeloid leukemia (AML), myelodysplastic syndromes (MDS), myeloproliferative neoplasms (MPN), myelodysplastic and myeloproliferative (MDS/MPN) neoplasms, and other rare diseases associated with eosinophilia and abnormalities of growth factor receptors derived from platelets or fibroblasts [1,2]. The classification of these diseases has been modified to include recent advances in cytogenetics and molecular genetics for hematologic malignancies.

In cancer registries, the International Classification of Diseases for Oncology (ICD-O) has been widely used since 1976, for coding the tumor site and histology. The most recent publication of ICD-O is the third edition (ICD-O-3), published in 2000 and revised in 2013 [3]. This updated version of ICD-O-3 includes the new terms and code changes suggested by the updated WHO classification on tumors of hematopoietic and lymphoid tissues in 2008. Recent cancer registry reports adopted this ICD-O-3 classification, as it was considered to better reflect our current understanding of diseases [4,5,6].

In Korea, the Ministry of Health and Welfare started the Korea Central Cancer Registry (KCCR), a nationwide hospital-based cancer registry in 1980. The KCCR expanded to include the entire population under the population-based cancer registry program since 1999 [7]. ICD-O-3 was applied to all incident cases for neoplastic diseases since the year 2003.

The KCCR and the Korean Society of Hematology (KSH) reported the first nationwide statistics of hematologic malignancies 4 years ago, covering the incident cases from 1999 to 2008 [8]. In the previous analysis, disease entities were defined based on the International Classification of Diseases 10^th edition (ICD-10) and calculated incidence and survival according to broad disease groups such as non-Hodgkin lymphoma (C82-C85, C96), myeloid leukemia (C92-C94), and lymphoid leukemia (C91).

The main objective of the current study is to update the statistical data on hematologic malignancies, focusing on myeloid malignancies, with the most recent database of KCCR in 2012. Incidence and survival estimates were analyzed according to more detailed disease groups compared to the previous report, using the ICD-O-3 codes. We also investigated recent epidemiologic changes of myeloid malignancies for decades in Korea with database from 1999 to 2012.

Incident cases of myeloid malignancies between 1999 and 2012 were obtained from the Korean National Cancer Incidence Database (KNCIDB) [9]. Myeloid malignancies were defined according to the revised version of ICD-O-3 (2013) and each code was categorized as shown in Supplementary Table 1, taking into account the clinical relevance. Myeloid malignancies were grouped into five categories: AML, MPN, MDS, MDS/MPN, and unknown myeloid neoplasms. AML and MPN were further divided into two subgroups as shown in Table 1, in consideration with acute promyelocytic leukemia (APL) and chronic myelogenous leukemia (CML), those have unique characteristics, treatments, and clinical outcomes separated from non-APL AML and non-CML MPN. For MPN, MDS, and MDS/MPN, incident cases between 1999 and 2002 were not included in this analysis because official registration by KCCR employing ICD-O-3 began in 2003. Codes for these categories by ICD-O-3 had significantly changed, but it could not be converted from old data.

Crude incidence rates (CRs) and age-specific incidence rates of each myeloid malignancy were calculated. The CRs per 100,000, an incidence rate based on the frequency of cancer in the entire population, were calculated by dividing the total number of events (N) by the total number of person-year of observation (P) and multiplying the result by 100,000. The age-specific incidence rates per 100,000 within age group i, were calculated by dividing the number of incidence observed in the age group (N_i) by the number of corresponding person-year of observation (P_i) and multiplying the result by 100,000. Age-standardized incidence rates (ASRs), weighted average of crude age-specific rates, were determined by using the mid-year population in Korea in 2000 as the standard population. Both age-specific incidence rates and ASRs were calculated according to age group: <14, 15-34, 35-49, 50-64, 65-79, and ≥80 years. Changes in the annual ASRs were examined by calculating the annual percentage change (APC) over a time period as (exp(b)-1)×100, where b is the slope of the regression of log(ASR) on a calendar year using the following linear regression equation [10]: E(log(ASR)|year)=a+byear

For the survival analysis, five-year relative survival rates were calculated. To find any changes for recent decades, KNCIDB data from 1996 to 2012 were included. The survival status of these patients was followed until December 31, 2013. Five-year relative survival (RS) rates for AML, and unknown myeloid neoplasms were calculated according to four periods of diagnosis: 1996-2000, 2001-2005, 2006-2010, and 2008-2012. RS for MPN, MDS, and MDS/MPN were calculated according to three periods of diagnosis (2001-2005, 2006-2010, and 2008-2012) since data before 2003 for these entities were not reliable. RS rates were developed by comparing the observed survival in the cancer patient group with the expected survival of the general population [11]. Five-year RS rates using the Ederer II method were based on an algorithm written by Paul Dickman in SAS [12,13]. All analyses were performed using SAS version 9.2.

We showed the epidemiologic data for myeloid malignancies in Korea between 1996 and 2012, updating the previous report [8], which had focused on analyzing the basic characteristics and past survival rates of domestic hematologic malignancies roughly classified according to ICD-10 code [14]. In this study, we adopted the latest ICD-O-3 codes [3] and the WHO classifications [1] to distinguish many different subcategories based on the cell lineage, histologic/genetic characteristics, and clinical prognosis, to avoid grouping clinically different diseases together as a single vague entity such as 'myeloid leukemia'. Analysis with subcategories enriched this study with data closely related to real-life practice, although it caused some artifacts in survival estimates owing to the small number of rare diseases.

Calculating population-based relative survival can measure survival improvement in each myeloid malignancy, which can be more easy compared to conventional cohort approaches or clinical trials. In Korea, the KNCIDB includes entire annual population diagnosed with malignant diseases, which is enough to truly represent the nationwide status.

Consistent with the first report by KCCR [8], we found an increasing incidence of most myeloid malignancies by age and year at diagnosis. Furthermore, increasing incidences in MDS, MPN, MDS/MPN were also detected, which could not be calculated in the previous report. Age and environmental factors [15,16,17] can be considered a potential cause of inflated incidence. A similar change was observed in the European HAEMACARE project [18], but Korean ASRs of MDS and MPN are much lower than that of European, probably because of ethnic or regional differences. The increased number of cancer survivors owing to the improvements in overall cancer survival [19] also needs to be considered because the risk of myeloid malignancies increases after chemoor radiotherapy [20], although the exact effect would be difficult to validate.

The relative survival for most myeloid malignancies was comparable to other country. Five-year survival of patients with AML (non-APL AML) was 29.9% in 2008-2012 and this is better than or comparable to European data (15% in 2006-2008) [21] and US data (25.9% in 2005-2011) [6]. Because the conventional chemotherapy with anthracycline and cytarabine has been widely applied for AML without significant change for years, prolonged survival may be due to the improvement in transplantation and supportive care such as anti-infective agents and immunosuppressive agents [22,23,24]. Best supportive care might influence on reduced early mortality in APL, with an introduction of arsenic trioxide for relapsed disease or frail patients to anthracycline.

For MDS, relative survival was significantly different between the subtypes, ranging from 24.3% for refractory anemia with excess blasts to 68.1% for MDS with 5q deletion. This is the first time that survival analysis according to the subtypes of MDS was performed.

Poor survival of patients with AML and high-risk MDS was more apparent in patients over 65 years old. It is well known that elderly AML or MDS patients are associated with unfavorable tumor biology compared to younger patients, but they also tend to be too frail to undergo intensive treatments, including hematopoietic stem cell transplantation required to induce the potential cure [25,26]. Survival improvements in elderly AML and MDS were not apparent compared to younger patients so far. In Korea, hypomethylating agents were approved for elderly patients with AML [27]. We should find the effect of this noble treatment in the future analysis

In the presented data, relative survival of CML strikingly increased in the 2000s, and became more apparent in the late 2000s. The main cause of such improvement is attributed to the introduction of tyrosine kinase inhibitors, as expected. In Korea, imatinib, the first tyrosine kinase inhibitor, was introduced in 2001 as a phase III clinical trial for approval. After its formal approval in 2006, this effective treatment has been prescribed widely. Five-year relative survival for patients with CML in our data (69.4% in 2001-2005 and 83.7% in 2006-2010) seems to be higher than in data from western countries (SEER database: 63.2% in 2005-2011 and EUROCARE data: 46% in 2003-2005) [6,21], although direct comparison between the databases has many limitations.

In terms of unknown myeloid malignancies, acute leukemia with ambiguous lineage and other vague entities (leukemia, acute leukemia, and myeloid leukemia) that cannot be classified to any categories of the WHO classification. The proportion of cases reported with vague terms among all myeloid malignancies has decreased from 24.2% in 1999 to 8.6% in 2012. To reduce this proportion further, we should make an effort to diagnose the disease using exact terms and codes. Annual education and update of recent changes in classification of hematologic diseases should be provided for medical record administrators.

In summary, we found an increasing incidence according to age and year at diagnosis for most myeloid malignancies. Patient survival was also improved over the study periods, mainly because of a notable increase in survival of patients with APL and CML owing to the widespread use of more effective and less toxic treatments and supportive care. Serial reports of population-based cancer registry data enable us to monitor the improvements in survival associated with recent advances in diagnosis and treatment for hematologic malignancies. To perform this essential task, maintaining the quality of cancer registries through comprehensive data collection and monitoring, adequate education programs, and recruiting stable investment for resources are prerequisites. A qualified registry would provide more amount of informative evidence as to whether the recent changes in treatment do lead to improved survival.

Incident cases of myeloid malignancies in men (B) and women (C) in Korea, 2012. ^a)AML cases include 153 cases of APL in both genders (77 cases in men, and 76 cases in women); ^b)MPN cases include 458 cases of CML in both genders (288 cases in men, and 170 cases in women); ^c)Unknown cases include 15 cases of acute leukemia, ambiguous lineage in both genders (7 cases in men, 8 cases in women).

Abbreviations: AML, acute myeloid leukemia; MPN, myeloproliferative neoplasms; MDS, myelodysplastic syndrome; MDS/MPN, myelodysplastic/myeloproliferative neoplasms.

Incident cases of myeloid malignancies by age group in Korea, 2012.

Abbreviations: same as in .

Age-specific incidence rates of myeloid malignancies in men (A) and in women (B) in Korea, 2012.

Abbreviations: same as in .

Trend in relative survival rate of myeloid malignances between 1996 and 2010 in Korea.

Abbreviations: same as in .