The epidemiology of haematological cancers in Sarawak, Malaysia (1996 to 2015)

The data was obtained from SCR managed by Sarawak State Health Department. The source of SCR was primarily based on passive voluntary notifications from all sections of the medical profession in Sarawak using a standardised format, called Cancer Notification Form (CNF) (Supplementary Data – Cancer Notification Form). In addition to that, active case finding and routine checks on government and private hospital discharge records and pathology record listings from the government and private laboratories were performed to ensure completeness of the data. This was supplemented by case detection from death records supplied by the National Registration Department. The cases picked up were then checked against data in SCR. If they were not yet notified, reminders were sent to the doctor-in-charge for more information. Additional missing data for these cases were sought by registry staff from the relevant hospital records.

Data were managed using the CanReg5 software from the International Association of Cancer Registries. The issue of multiple notifications of a patient was addressed by cross checking names and National Identification Card (IC) numbers, which are unique to each Malaysian citizen and permanent residents (PR), using the software.

This study was registered with the National Medical Research Register and approved by the Medical Research & Ethics Committee of the Ministry of Health, Malaysia (NMRR ID-22–01151-6KI (IIR)).

Descriptive analysis was performed using Excel for Microsoft 365. Multivariable analysis of incidence rate (c.f. Incidence Rate (IR)) was analysed with negative binomial regression analysis using Stata version 9.0 software (StataCorp, College Station, TX) and adjusted to gender, ethnic, and age groups. The result was reported as incidence rate ratio (IRR) and 95% confident interval (CI). The IRR is similar to risk ratio with the difference of including time variable in the calculation of incidence. The ratio compares the incidence of cancer in the target group of subjects against the reference group. Ratio of more than 1 indicates higher incidence in comparison to the reference group. The reference group used for the calculation of IRR are female, Malay, and age group of 0 to 4 year-old for gender, ethnic, and age groups respectively. Cancer trend analysis was conducted using Joinpoint Regression Program version 4.9.10 from the National Cancer Institute.

A total of 3,947 incidences were identified from SCR. CR of haematological cancers in Sarawak 1996 to 2015 was 8.9 per 100,000 population with male predominance against female (IRR 1.32, 95%CI 1.24,1.41) (Table 1). The lifetime risk of a male develops haematological cancer was 1.5 time higher than a female.

Incidence in all the ethnic groups is shown in Supplementary Data – Supplementary Table 2. Bidayuh and Melanau had the highest and lowest CR among the five major ethnic groups, 10.9 and 5.3 per 100,000 population, respectively (missing data n = 4) (Table 2). There was a difference among the five major ethnic groups in comparison to Malay – Bidayuh and Melanau had the highest (IRR 1.13, 95%CI 1.00, 1.27) and lowest (IRR 0.54, 95%CI 0.45, 0.65) incidence rate, respectively. Iban also had a different incidence rate (IRR 0.86, 95%CI 0.79, 0.94) compared to Malay. There was no difference between Chinese and Malay (IRR 0.93, 95%CI 0.85, 1.01). Crude Rate gender ratio was also different among the five ethnic groups with Bidayuh had the highest male predominance, 1.4, while Melanau had the lowest, 1.1.

After excluding 255 missing data, a total of 3,692 incidences were analysed according to the 11 administrative divisions (Table 3). There was a difference in CR and CR gender ratio among the administrative divisions.

Incidence Rate showed a peak at initial years of life (age 0 to 4) which reduced to the lowest at the age of 15 to 19 years, then slowly increased with age with exponential rise after the age of 40 years for both genders till the age of 65–69 years. The average of gender-specific IR male to female ratio was 1.3 with higher ratio of 1.6 in older age group of 60 years and above (Fig. 1).

The pattern of distribution of the gender- and ethnic-specific IR among the five major ethnic groups was similar both to each other and to that of the whole population (Fig. 1). In older age group of 60 years and above, gender-specific IR male to female ratio still showed male predominance in all four ethnic groups with the highest in Malay, 1.9, followed by Bidayuh, 1.8, while Melanau had an inverse ratio, 0.8.

As the population distribution across the 5-year age groups might be different among the five major ethnic groups, ASR is a better representation than CR which could be appreciated in Table 2. The descending order of ASR according to the ethnic groups was Bidayuh 12.7 per 100,000 population, Malay 10.8, Chinese 10.2, Iban 9.4, and Melanau 5.9. The highest ASR gender ratio was also different from CR gender ratio, of which Iban was the highest, 1.5, while the lowest was the same, Melanau, 1.1.

Incidence of specific haematological cancer categories is shown in Supplementary Data – Supplementary Table 3. Among the three major haematological cancer categories, namely lymphoma, acute leukaemia (AL), and plasma cell neoplasm (PCN), or more commonly known as lymphoma, leukaemia, and myeloma, the CR ratio was 8: 4.5: 1 (Table 4). Crude Rate of non-Hodgkin lymphoma (NHL)/leukaemia was 8 times higher than Hodgkin lymphoma (HL). Among NHL/leukaemia, CR of mature B cell was 7.5 times higher than T cell. Crude Rate of acute myeloid leukaemia and related precursor neoplasms (AML) which includes acute promyelocytic leukaemia (APML) was similar to precursor lymphoid neoplasms i.e., acute lymphoblastic leukaemia (ALL). Of note, the incidence of mature B cell NHL/leukaemia only began to increase since 2009 because immunohistochemical staining of sample has only become more widely available since then. Male predominance was present in all haematological cancer categories except AML and mature T cell NHL/leukaemia which had equal gender CR.

As the IR was lowest at the age group of 15 to 19 years (Fig. 1) and there was no standard definition of paediatric and adult ages, the cut-off age of 15 years was used to compare between paediatric and adult haematological cancers. Ages of 15 years and above was categorised as adult, while below as paediatric. There was a difference in CR ratio of the above specific haematological cancer categories between paediatric and adult (Table 4). In paediatric, PCN was extremely rare and CR ratio of lymphoma: AL was 1: 4. Crude Rate ratio of NHL: HL was 4: 1, mature B cell NHL: T cell NHL/leukaemia was 4: 1, and AML: ALL was 1: 2.5. In adult, CR ratio of lymphoma: AL: PCN was 7.5: 2.5: 1, NHL: HL was 7.5: 1, mature B cell NHL: T cell NHL/leukaemia was 8: 1, and AML: ALL was 3: 1. Haematological cancers were generally male predominant in both paediatric and adult, but certain cancer categories had higher male predominance in paediatric than adult i.e., HL, 2.1 versus 1.4, and mature B cell NHL/leukaemia, 1.8 versus 1.3.

AML, ALL, mature B cell NHL/leukaemia, HL, PCN, and chronic myeloid leukaemia (CML) were further analysed because the total incidence was more than 200. Summary of the incidence and CR of the six selected haematological cancers is shown in Supplementary Data – Supplementary Table 4. The pattern of gender-specific IR in AML, mature B cell NHL/leukaemia and PCN were similar, showing an inverted pyramidal shape with IR increased exponentially from the age of 40 (Fig. 2). On the other hand, ALL showed a pyramidal shape, with a linear decreasing trend across age groups between 0 to 19. IR showed a slow increasing trend across age groups in HL and CML. Incidence Rate of the six selected cancers was further breakdown according to the five major ethnic groups as shown in Fig. 3.

All the six selected cancers were male predominant, except AML which had an equal ASR gender ratio (Table 5). The highest male predominance was in CML, 1.6, and lowest in ALL, 1.3. The commonest cancer in male and female was mature B cell NHL/leukaemia, 1.7, and AML, 1.4, respectively.

There were differences in ASR of the six selected cancers among the five major ethnic groups (Table 5). In general, Melanau had the lowest ASR for all the six selected cancers, between 0.4 and 0.9 per 100,000 population. Bidayuh had the highest ASR for mature B cell NHL/leukaemia, HL, and PCN, 2.3, 1.0, and 1.1, respectively, compared to the lowest in Melanau, 0.9, 0.4, and 0.4, respectively. Chinese and Malay had the highest ASR for AML, 1.6, compared to the lowest 0.7 in Melanau. Chinese had the highest ASR for ALL, 1.7, while Bidayuh and Melanau had the lowest, 0.8. Malay had the highest CR for CML, 0.7, compared to the lowest, 0.4, in Melanau and Chinese. The ASR gender ratios were also different among the five major ethnic groups, but generally still male predominant. The highest ASR gender ratio was 10.7 in HL Melanau, while the lowest was 0.5 in AML Melanau. The loss of male predominance was seen more in Melanau, while Bidayuh showed male predominance in all the six selected cancers. Chinese showed loss of male predominance in AML and CML, Malay in AML and HL, and Iban in AML and mature B cell NHL/leukaemia.

In Bidayuh, the commonest among the six selected cancers was mature B cell NHL/leukaemia with ASR of 2.3, followed by AML, 1.4, while the rarest was CML, 0.5 (Table 5). In Chinese, the commonest was ALL, 1.7, followed by AML, 1.6, while the rarest was CML, 0.4. In Malay, the commonest was mature B cell NHL/leukaemia and AML, 1.6, followed by ALL, 1.4, while the rarest was HL and PCN, 0.6. In Iban, the commonest was mature B cell NHL/leukaemia, 1.3, followed by AML, 1.2, while the rarest was CML, 0.5. In Melanau, the commonest was mature B cell NHL/leukaemia, 0.9, followed by ALL, 0.8, while the rarest was HL, PCN and CML, 0.4.

In paediatric, the ASR gender ratio was generally more than adult except in ALL, 1.2 versus 1.5 (Table 5). The highest ASR gender ratio in paediatric was seen in HL, 2.1 with Iban had the highest, 3.0. ALL is the commonest haematological cancer in paediatric. However, there was an ethnic difference with the highest in Chinese, 1.2, followed by Malay, 1.0, while the lowest in Bidayuh and Melanau, 0.6. It appeared that Malay paediatric had a higher risk of getting AML as compared to other ethnic groups. It seems like Chinese had more haematological cancer involving lymphoid lineage during childhood but myeloid lineage disease increases in adulthood, while the other four ethnic groups continued to have more lymphoid lineage disease in adulthood.

In adult, the highest ASR gender ratio was 3.4 in CML Malay and lowest 0.4 in AML Melanau (Table 5).

From the available subtyping data, the two major subtypes of HL were nodular sclerosis (ICD-10: C811; ICD-O: 9663/3 – 9667/3) 29/52 (55.8%) and mixed cellularity (ICD-10: C812; ICD-O: 9652/3) 19 (36.5%).

The three major subtypes of mature B cell NHL/leukaemia were diffuse large B cell lymphoma (ICD-10: C833; ICD-O: 9680/3, 9684/3, 9688/3, 9735/3, 9737/3) 346/546 (63.4%), chronic lymphocytic leukaemia/small lymphocytic lymphoma (CLL/SLL) (ICD-10 and ICD-O for CLL: C911 and 9823/3; for SLL: nil (no specific coding) and 9670/3) 78 (14.3%), and follicular lymphoma (ICD-10: C820 – C829; ICD-O: 9695/3, 9691/3, 9698/3, 9597/3, 9675/3, 9690/3) 66 (12.1%).

The two major subtypes of mature T cell NHL/leukaemia were peripheral T-cell lymphoma, not elsewhere classified (ICD-10: C844; ICD-O: 9720/3) 31/75 (41.3%) and anaplastic large cell lymphoma, ALK-positive (ICD-10: C846; ICD-O: 9714/3) 23 (30.7%).

Among the specific subtypes of AML, APML (ICD-10: C924; ICD-O: 98,660/3) consisted of 30/596 (5.0%), acute erythroid leukaemia (ICD-10: C940; ICD-O: 9840/3) 5 (0.8%), and acute megakaryoblastic leukaemia (ICD-10: C942; ICD-O: 9910/3, 9911/3) 2 (0.3%). Among ALL, B cell (ICD-10: nil; ICD-O: 9836/3) was 27/34 (79%) and T cell (ICD-10: nil; ICD-O: 9837/3) was 7 (21%).

Of note, the discussion in this sub-section should be interpreted together with the limitation of the study discussed in the sub-section below.

The estimated average percentage change in Sarawak from 1996 to 2015 for AML, ALL, CML, and CLL in comparison with other countries [12] is shown in Table 7.

The decreasing trend of ASR in AML was contradicting with the increasing trend observed in most countries between 1990 and 2017 [12].

For ALL, the same decreasing trend was observed in most countries between 1990 and 2017 [12]. Interestingly, the decreasing trend in South Korea [12] is contradicting with another study which showed an increasing trend in lymphoid leukaemia between 1999 to 2018 [13]. This might be due to increasing trend of lymphoid leukaemia other than ALL. However, the decreasing trend in Spain [12] is consistent with another study between 1995 to 2015 [14].

For CML, the CR of 0.5 per 100,000 population in Sarawak between 1995 to 2015 was similar to the previously reported 0.8 in southern Sarawak between 2011 to 2016 [15]. The decreasing trend was observed in most countries between 1990 and 2017 [12]. The decreasing trend in South Korea [12] is, again, not shown by another study which showed an increasing trend in myeloid leukaemia between 1999 to 2018 [13]. This is most likely masked by the increasing trend of AML, a much commoner myeloid leukaemia than CML.

PCN is a haematological cancer category with a known high incidence rate in elderly, of which the same observation was seen in Sarawak population among those aged 60 years and above (Fig. 2 (e)). With the aging population and better sociodemographic index, the decreasing trend in PCN was contradicting with the expected increment seen in other studies using global data from 1990 to 2016 [16].

For NHL with ICD-10 coded C82-86, the increasing ASR trend in Sarawak 1996 to 2015 was consistent with most countries [17].

Like most registry data, this study is also limited by under-reporting. The active case finding and routine checks had reduced the under-reporting rate, but a significant number was likely persisted as evidenced by only 4 incidence of PCN in 2015 which is very unlikely based on the authors’ clinical experience and wave-like pattern in the overall incidence over the 20 years (Fig. 4). To our postulation, the under-reporting could be due to, firstly, a limited access to health care system in the remote areas in Sarawak between 1996 and 2015 that led to under-diagnosed cases. However, this would reduce as Sarawak develops. Secondly, the data source in the study did not include death record registry to capture unnotified cases that might escape active case finding and routine checks. Thirdly, changes in personnel in the system might led to inconsistent workflow in diagnosing, notifying, and finding/checking. However, this could be reduced by including data over a longer period of time.

MDS and some diseases in the specific haematological cancer categories (Supplementary Data – Supplementary Table 3) are important haematological neoplasms which contribute significant workload to the daily clinical haematological practice. However, they were not included in the analysis (see MATERIALS AND METHODS) because they are likely under-diagnosed and under-reported. For example, to diagnose MDS in the presence of cytopaenia, an invasive procedure involving bone marrow aspiration and trephine biopsy is required to assess for dysplasia and genetic abnormalities. However, in an asymptomatic elderly patient with mild cytopaenia not caused by other identifiable causes, the procedure might be refused or delayed by the treating physician and/or patient who opted for blood count monitoring only. Thus, this study was confined to haematological cancers with ICD-10 coded C81-C96.

There are many classifications of haematological neoplasms, which evolve and/or merge over time as our understanding of these diseases advances. The World Health Organization (WHO) Classification of Tumours of Haematopoietic and Lymphoid Tissues is the reference classification used in daily clinical practice, and it is updated every few years. The latest version is the recently released 5^th edition [18, 19]. The evolution of the classification of haematological neoplasms must be considered during data interpretation of the study. Firstly, the study was unable to classify 1,277 cases of NHL/lymphoid leukaemia, not otherwise specified (NOS), 11 cases of myeloid leukaemia, NOS, and 46 cases of leukaemia, NOS (Supplementary Data – Supplementary Table 3) into more specific categories/diseases due to the evolution of the classification of haematological neoplasms over time and logistic constraint. Secondly, the incidence of mature B or T cell NHL/leukaemia was unavailable prior to 2009 because immunohistochemical staining of samples has only become more widely available in Sarawak since then. Thirdly, the study could not provide information on certain important specific subtypes for daily clinical haematological practice like ALL with BCR::ABL1 fusion, and AML with RUNX1::RUNX1T1 fusion and CBFB::MYH11 fusion. AML with CBFB::MYH11 fusion is not specified in both ICD-10 and ICD-O, while ALL with BCR::ABL1 fusion and AML with RUNX1::RUNX1T1 fusion are not specified in ICD-10 but specified in ICD-O, code: 9812/3 and 9896/3, respectively. There was no cases of ALL with BCR::ABL1 fusion and 15 cases of AML with RUNX1::RUNX1T1 fusion in the study which does not correspond to common epidemiology and authors’ clinical experience. Regardless of specification in ICD-10 or ICD-O, specific subtypes were likely under-reported because result of the specific genetic tests are usually delayed and unavailable at the time of diagnosis and reporting of ALL or AML.