To our knowledge, this is the first comprehensive report on the burden of hematological malignancies in Bangladesh. In contrast to the WHO estimates, our multi-centred hospital-based data present a different picture: the leukemias constituted approximately two thirds of (64.3%) all HM cases, while NHL accounted for 16.9%, followed by MM (10.5%) and HL (3.9%) [Table 2]. A similar pattern of leukemias (Age-standardized incidence rate or ASR per 100,000 is 3.3), NHL (ASR per 100,000 is 3), and multiple myeloma is observed in India. The WHO prediction, the commonest type of HM was NHL (ASR is 1.9 per 100,000 persons), which was followed by leukemias (ASR 1.7 per 100,000 persons), HL and multiple myeloma . In Pakistan, NHL is the most prevalent type of HM . In US, NHL is the commonest cancer among HM, which is 1.5 times that of all leukemias . In contrast, all leukemia cases were over three times higher as compared to NHL cases in Bangladesh. In other Asian countries including Japan, Korea and Singapore, NHL is the most frequent hematological malignancies [1, 7]. In our study, this unexpected discrepancy might be due to lack of proper referral system in some participating centres. Although lymphoma is a haematological disorder, a small number of patients might have been admitted to the medical oncology department. Moreover, year-specific data was not available for four of the participating centres (Table 1). Yet, our hospital-based study represents the overall current country picture on HM. Additional population-based studies are warranted to reveal the true incidence of HM.
Unlike Western countries, the hematological malignancies in Bangladesh seem to afflict younger population as is indicated by the overall median age at diagnosis was 42 years. Perhaps the true median age may be even lower, given the under-representation of children in this study. Acute leukemias (AML and ALL), CML and lymphoma (NHL and HL) were found to occur in relatively young adults with a median age ranging from 27 to 48 years (Table 4). On the other hand, MM, CLL and MDS were associated with relatively older individuals (median age at diagnosis was 55 to 60 years). It is generally argued that young age phenomenon of cancers might be due to the lower life expectancy and younger population structure of a respective country [12, 15, 16]. In Bangladesh, 33% of the population is under 15 years old and the median age of the entire population is 23.4 years. Another often cited reason is related to the underreporting cases of older individuals possibly because of several socioeconomic and cultural reasons. However, the lower life expectancy explanation is doubtful as in economically developed Singapore, Hong Kong, South Korea, Taiwan and Japan - with much higher life expectancies similar to those seen in Western countries - are also affected by CML , NHL  and MDS  at relatively young ages. The reasons behind this phenomenon are unclear. However, it is likely that the multiple factors including genetic, infections and other environmental factors might play crucial role in this young age phenomenon in Asia. Moreover, an analysis with migrant Bangladesh/Indian living in economically developed regions with different life expectancy may help to understand this issue further.
Our retrospective analysis included 6.7% childhood (under 20 years) HM cases (n = 338), of which approximately 76% cases comprised of adolescent (15–19 years age group) cancer patients. It is important to mention that childhood cases were under represented in the present study as the information was obtained from the hematology department of some of the major participating centres which usually manage childhood (0–14 years) cases in separate facilities. Despite this limitation, we can present an overview of children hematological malignancies in Bangladesh. Leukemias were found to be the most common blood associated cancers, constituting of 92.6% of all children HM cases (Table 3). Among these leukemia patients, ALL is the most frequent cancer accounting for 37.3% (n = 126) followed by AML (34%, n = 115) and CML (13%, n = 44). Lymphoma accounted for 13.6% (HL-8.3% and NHL-5.3%) of all children cases. Like Western countries and Bangladesh, the leukemias were the most prevalent childhood cancers in India with a relative proportion ranging from 25-40% of all childhood cancers. Approximately 60 to 85% of all reported leukemias was ALL . Further extensive study is necessary to understand the pattern and distribution of HM among children in the country.
Studies all over the world have revealed that hematological malignancies is gender-skewed, often affecting men more than women. We also found that men were more involved than women, with an overall male to female ratio of 2.2. For lymphoma, this ratio (~3.4) seems to be much higher than normal trend. Probably, female cases were underreported considering the socioeconomic status of the families and in low-resource settings usually men often get priority while seeking medical attention. The higher prevalence of HM in males might be the result of increased exposure to environmental and occupational risk factors, smoking, alcohol consumption as well as different hormonal and genetic background of males and females [10, 21–23].
Acute leukemias including AML and ALL are the most prevalent HM affecting Bangladeshi population, accounting for 42.4% (n = 5013) of all HM cases, while these two constituted 66% (n = 3218) of leukemia cases (Table 2). The frequency of AML is two times higher than that of ALL in Bangladesh. The incidence of AML is relatively common in North America, Europe, and Oceania, while adult AML is rare in Asia and Latin America . AML generally affects older individuals with a median age at presentation of around 65 years in Western countries  and it accounts for ~ 29% of all leukemias in adults in US. Our study showed that acute leukemias tend to affect relatively young adults aged 20–49 years (66.4% cases) [Table 3]. The median age at onset for AML (35 years) in Bangladesh is higher than in India (30 years) . Apart from lower life expectancy prevailing in the Indian subcontinent, it is likely that elderly patients may not be reporting to the hospitals because of relatively rapid progression of AML. ALL occurs in people of all ages but it exhibits bimodal age-specific curve with peaks in youngest (<20 years) and oldest ages (>50 years). In our study we observed that more than 55% of the ALL occurs in young adults (20–40 years) with a median of 27 years (Table 3). In US, the overall median age for ALL is only 14 years, since approximately 60% of the cases occur in children under 20 years old . In case of adult, the median age was 38 years .
Chronic leukemias constituted 21.9% of all HM and 34% of leukemias in Bangladesh. CML is the second most common type (18.2%, n = 912), while CLL is the least frequent (3.7%, n = 183) HM (Table 2). The frequency of CML is five times higher than CLL. The pattern of CML occurrence is different in India and Africa where it is the most common form of leukemia [12, 27]. In Sudan, the incidence of CML is very high, being the predominant cancer in men in last 25 years . CLL is a rare hematological malignancy in Asia while this is the commonest form of leukemia affecting elderly in Western countries with a median age of 70 years  . In US, for instance, CLL constitutes about ~34% of all leukemia . In Bangladesh and India, CLL is found to occur mainly in adults (median age ~60 years) . The lower frequency of CLL in Bangladesh and its neighbouring countries may not be associated with lower life expectancy since the incidence of CLL in Japan is at least 4 to 5 times lower than Western countries; despite Japanese have the highest life expectancy in the world . It is possible that genetics and environment may play important role in its development [11, 31].
Malignant lymphoma constituted 20.8% of all HMs in our study (Table 2). Out of 1052 lymphoma cases, NHL accounted for 80.4% while 19.6% was for HL. The frequency of NHL and HL observed in the present study was almost similar to the earlier report from India . NHL is one of the commonest cancers in developed countries, but the incidence is relatively lower in Asia . This is the most frequent HM of many Asian [1, 18] and African countries . In Bangladesh, it constituted 16.9% of all HM with a median age of 48 years. On the other hand, HL is the most common cancer of young adults in developed countries. The age distribution of HL is bimodal, the first being young adults (age 15–35 years) and the second being those in older individuals over 55 years old. In our study, 62.7% of all HL cases occurred among young adults (<50 years) with a median of 36 years. The similar frequency of HL pattern was also observed in India where mean age of diagnosis was 31.9 years .
In our study, nearly 76.7% of the MM patients aged over 50 years. This accounted for 10.5% of all HM. In Western countries, the median age at diagnosis is 65–70 years, which is significantly higher as compared to Asian countries like India, Japan and China . The median age at onset in Bangladesh is 55 years, which similar to that in India (55–56 years). The incidence varies among different races or geographic location. For instance, the higher incidence has been reported for African Americans as compared to Caucasians .
Myelodysplastic syndromes (MDS) are relatively a condition affecting elderly. It can transform to AML. The incidence of MDS is unclear worldwide because of historical lack of population-based registration. In Bangladesh, MDS constituted approximately 4.5% of HM and 76.9% (n = 225) of the patients were over 50 years old with a median age of 57 years, which is similar to that reported in Japan . In case of India, the MDS patients were much younger with a mean age of 46.1 years , while Western countries have median age ranges from 60–70 years .
Large sample size (over 5000 diagnosed cases) is the strength of our study. However, in our retrospective study, we could not provide any information on the mortality and survivorship of HM, since there is no proper follow-up system to track terminally-ill patients in Bangladeshi hospitals. Detailed clinical data were also not available because of incomplete record keeping system. Like other low-income countries, many of these limitations are intrinsic to Bangladesh health system. Time-trend analysis was not also possible due to lack of homogeneity of data obtained from different hospitals, which has been described in the method section and elsewhere.