Thalassaemia and risk of cancer: a population-based cohort study.

Thalassaemia and risk of cancer: a population-based

cohort study

Wei-Sheng Chung, ^1,2 Chun-Liang Lin, ¹ Cheng-Li Lin, ^3,4 Chia-Hung Kao ^5,6

INTRODUCTION

Thalassaemia is one of the most common autosomal recessive diseases worldwide.

Thalassaemias are inherited haemoglobin (Hb) disorders caused by reduced or absent production of

one or more of the globin chains that comprise the Hb tetramers. They are prevalent in the

Mediterranean area, Middle East and Southeast Asia.

The overall prevalence rate of the thalassaemia trait is approximately 4.5% in Taiwan.

Thalassaemia may cause chronic microcytic anaemia and haemolytic anaemia, depending on the disease severity. People with thalassaemia can accumulate an iron overload in their bodies, either from the disease or from frequent blood transfusions.

Infections are a frequent complication in

patients with thalassaemia.

Predisposing factors for infections in patients with thalassaemia may include severe anaemia, iron overload, splenectomy and a range of immune abnormalities.

Previous studies have indicated that iron overload is associated with colorectal cancer and hepatic cancer.

Infections with hepatitis B or C viruses also increase the risk of hepatic cancer.

Lymphomas existing in patients with thalassaemia were reported by previous studies.

Currently,

few studies have explored the epidemiological relationship

between thalassaemia and risk of

cancer.

Therefore, we conducted a nationwide population-based cohort study to investigate whether patients with thalassaemia have an increased risk of cancer.

METHODS Data source

This study used the Longitudinal Health Insurance Database 2000 (LHID 2000) released by the Taiwan National Health Research Institutes (NHRI) for research purposes. The National Health Insurance (NHI) programme was launched in Taiwan in 1995, which covers 99% of the Taiwan population of 23.74 million.

The LHID 2000 information consisted of one million benefciaries randomly sampled from the original NHI benefciaries.

The database includes the entire registry

and claims data from this health insurance system, ranging from demographic data to detailed orders from ambulatory and inpatient care. The privacy of patients and healthcare providers is protected in that researchers using the database are required to submit their study protocols for evaluation by the NHRI. This study was approved by the

Institutional Review Board of China Medical

University, Taiwan (CMU-REC-101-012). The diseases were coded according to the International

Classifcation of Disease, 9th Revision, Clinical Modifcation (ICD-9-CM) diagnosis codes, 2001 edition. The high accuracy and validity of ICD-9-CM diagnoses in the LHID have been described in previous studies.

Study population

On the basis of the LHID, enrollees with a thalassaemia (ICD-9-CM 282.4) diagnosis from 1

January 1998 to 31 December 2010 were included in the thalassaemia cohort. The frst date of thalassaemia diagnosis was defned as the index date.

Patients with a history of any type of cancer

(ICD-9-CM 140-208) or sickle cell/aplastic/haemolytic/

sideroblastic anaemia (ICD-9-CM

282.6-285.8) before the index date and those with incomplete age or sex information were excluded.

The sample size of the comparison cohort was fourfold that of the thalassaemia cohort with age (per 5-year

span) and gender frequency matching; patients were selected from insured residents without a history of thalassaemia, sickle cell/aplastic/haemolytic/sideroblastic anaemia and any type of cancer.

Outcome measurement

Both the thalassaemia and comparison cohorts were followed until the end of 2011 to estimate incident cases with the diagnosis of cancer. We obtained data on patients who were diagnosed with cancer from 1 January 1998 to 31 December 2011 from the Registry of Catastrophic Illness Patient Database (RCIPD), which is a subset of the NHIRD. Any diagnosis of cancer, except metastatic cancer (ICD-9-CM codes 140–195 and 200–

208), was coded by doctors and reviewed by offcials of the NHI. We divided cancer into the following eight groups: head and neck cancer (ICD-9-CM codes 140–149 and 161), cancer in abdomen except for the genitourinary (GU) system (ICD-9-CM codes151–159), cancer out of abdomen

(ICD-9-CM codes 60, 162–165, 170–174, 176 and 190–194), uterine cancer (ICD-9-CM codes 180–184), prostate cancer (ICD-9-CM code 185), urinary system cancer (ICD-9-CM codes 188–189), haematological cancer (ICD-9-CM codes 200–208) and other cancer types.

Variables of exposure

We also assessed patients who had at least three claims for ambulatory care visits or hospitalisation visits at the baseline with principal/secondary diagnoses of the following diseases including: anaemia (ICD-9-CM codes 285 and 286), cirrhosis (ICD-9-CM code 571), heart failure (ICD-9-CM codes 428 and 429.3), diabetes (ICD-9-CM code 250), viral hepatitis

(ICD-9-CM code 070) and blood transfusions (ICD-9 procedure

code 990), which were all recognised from the LHID 2000

as baseline comorbidities. Furthermore, patients receiving blood

transfusions (ICD-9 procedure code 990) based on thalassaemia severity were identifed according to their diagnoses in the study period.

Statistical analysis

The distribution of demographic status and comorbidities was

expressed as a frequency (with per cent) or mean±SD. The differences among continuous variables were estimated using t

tests, and the differences among categorised variables were analysed using the χ

test. The incidence densities of cancer were

calculated by sex, age and comorbidity for each cohort.

Univariable and multivariable Cox proportional hazard regression models were used to assess the risk of cancer associated

with thalassaemia, compared to the comparison cohort. HR and 95% CI were estimated in the Cox model. Multivariable models were adjusted for age, sex and comorbidities of anaemia, cirrhosis, heart failure and diabetes. We used the Kaplan-Meier

method to compare the cumulative incidence of cancer events between the two cohorts and used the log-rank test to examine the differences. SAS V.9.1 (SAS Institute, Cary, North Carolina, USA) was used for data analyses. The p value <0.05 was considered statistically signifcant.

RESULTS

Demographic characteristics and comorbidities of study participants

We identifed 2655 patients with thalassaemia as the thalassaemia cohort, and 10 620 patients without thalassaemia as the

comparison cohort between 1998 and 2010 (table 1). Women dominated the study cohorts (62.3%) and approximately 56.1%

of participants were aged less than 34 years. No signifcant differences in distribution of sex and age between the thalassaemia

and comparison cohorts were found (mean age 34.6±18.5 years vs 34.5±18.6 years, respectively). Patients with thalassaemia were more likely to have anaemia, cirrhosis, heart failure, diabetes, viral hepatitis and transfusion than were the comparison

cohort (all p<0.05). The mean follow-up years were 7.52

(SD=3.60) and 7.45 (SD=3.60) for the thalassaemia and comparison cohorts, respectively (data not shown).

Incidence density rates and HR for cancer according to the

thalassaemia status stratifed by demographic characteristics and comorbidity

Between the thalassaemia and comparison cohorts, the incidence density rates of cancers were 3.96 and 2.60/1000 personyears, respectively (table 2). The overall incidence of cancer was 52% higher in the thalassaemia cohort than in the comparison cohort, with an adjusted HR (aHR) of 1.54 (95% CI 1.15 to 2.07). The incidence of cancer was increased with age in both cohorts. Stratifed by sex, women with thalassaemia had a 70%

increased cancer risk compared to the comparison cohort. The age-specifc and comorbidity-specifc analyses failed to demonstrate a signifcantly higher aHR of cancer in patients with thalassaemia in any age subgroup, with or without comorbidity,

compared with their counterparts in the comparison cohort.

Risks of specifc cancer between the thalassaemia cohort and comparison cohort

Table 3 shows the specifc analyses of cancer types between the thalassaemia and comparison cohorts. Patients with thalassaemia exhibited a 5.32-fold increased risk (95% CI 2.18 to 13.0)

for haematological malignancy, and a 1.96-fold increased risk (95% CI 1.22 to 3.15) for cancer of the abdomen, except for the GU system, compared with those of the comparison cohort. Incidence rates and HR for cancer risk according to thalassaemia severity

We calculated the aHR according to the transfusion status in the study period after thalassaemia was diagnosed, as shown in table 4.

Compared to the patients with thalassaemia without transfusion, patients with thalassaemia with transfusion were 9.31-fold more likely to develop haematological malignancy (95% CI 1.46 to

59.3). Relative to the patients with thalassaemia without transfusion, those with thalassaemia with transfusion had a higher risk of

overall cancer (aHR=6.70, 95% CI 3.29 to 13.6) and cancer of the abdomen, except for the GU system (aHR=9.12, 95% CI 3.09 to 27.0).

Cumulative incidence of cancers, cancer in abdomen except the GU system, and haematological malignancy for the thalassaemia and comparison cohorts

Figures 1A–C show that the thalassaemia cohort had a signifcantly

higher cumulative incidence of overall cancer (p=0.001) (fgure 1A), cancer of the abdomen, except the GU system (p=0.004) (fgure 1B), and haematological malignancy (p<0.001) (fgure 1C) compared with the comparison cohort.

DISCUSSION

This nationwide study assessed the risk of cancer development among patients with thalassaemia in a large population-based cohort for an Asian population during a 14-year follow-up

period. The cancer incidence was higher in patients with thalassaemia than in the comparison cohort (5.18 vs 3.61/1000

person-year). Patients with thalassaemia exhibited a higher prevalence of comorbidities, namely anaemia, liver cirrhosis, heart failure and diabetes mellitus, than did the comparison cohort. However, patients with thalassaemia had a 1.47-fold overall risk of cancer compared with that of the comparison cohort after we adjusted for age, sex and comorbidities.

The actual mechanism of cancer development among patients with thalassaemia remains unclear. Patients with thalassaemia may be exposed to transfusion-transmitted infections caused by bacteria, parasites and viruses. The Epstein-Barr virus in patients with β-thalassaemia may produce chronic antigenic stimulations, which lead to the development of IgE, anti-IgG, anti-IgA and anti-leucocyte antibodies as a culprit for developing lymphoma production.

Human T-cell lymphotropicvirus-1 has also been reported to play a role in the pathogenesis of adult T-cell

lymphoma/leukaemia.

Hepcidin is primarily produced in the liver and is a major regulator of iron homeostasis in the body. Hepcidin controls

dietary iron absorption, plasma iron concentrations and tissue

iron distribution.

Hepcidin defciency is the primary or contributing factor of iron overload in patients with thalassaemia.

Iron overload is the principal cause of morbidity and mortality in β-thalassaemia with or without transfusion dependence. The non-transferrin-bound fraction of plasma iron might promote the generation of reactive oxygen species and peroxidative tissue injury.

Iron overload is a risk of cancer development.

The substantial increase in specifc cancer risk among the thalassaemia cohort included haematological malignancy

(aHR=5.32) and cancer of the abdomen (aHR=1.96) after we

adjusted for covariates. Furthermore, patients with thalassaemia who received blood transfusions carried a greater risk of developing haematological malignancy and abdominal cancer than

did the patients with thalassaemia without blood transfusion.

For severe thalassaemia, iron overload may result from multiple transfusions in addition to increased iron absorption and hepcidin suppression. Transfusional hemosiderosis is also a frequent

complication in patients with transfusion-dependent chronic diseases such as thalassaemias and haemoglobulinopathy.

Iron

overloading may cause end organ damage, endocrine dysfunction and bone deformity.

Furthermore, iron overload and

transfusion-related infection may additively induce carcinogenesis.

Certain limitations must be considered when interpreting the fndings of this study. First, most of the patients diagnosed with thalassaemia were aged 20 years and older in the study, which implied less severity of the thalassaemia and essentially excluded most major thalassaemia cases. Second, thalassaemia cases may be underestimated in the LHID because patients with the thalassaemia trait may present as silent carriers. This phenomenon might

result in underestimating cancer risk for patients with thalassaemia.

Third, the LHID does not contain detailed personal information, such as data on lifestyle habits and family histories of

malignancy; this information might have constituted a potential

confounding factor in this study. Fourth, a surveillance bias may patients with thalassaemia. Finally, we were unable to retrieve relevant

clinical information, such as haemoglobin levels, various types of thalassaemia (α-thalassaemia, β-thalassaemia and

δ-thalassaemia) and genetic information, because the identifcation numbers of the patients were scrambled.

The strength of this study is that it was a longitudinal nationwide

population-based cohort study investigating the cancer risk

in patients with thalassaemia. The NHI programme is universal

and mandatory in Taiwan, and NHI benefciaries are assigned

personal identifcation numbers that enable them to be traced

throughout the follow-up period. Moreover, cancer diagnosis of

the patients in our study was verifed based on their inclusion in

the RCIPD, indicating that the data were highly reliable.

In summary, this nationwide study of 2655 patients with thalassaemia with 19 960 person-years of follow-up indicated that

patients with thalassaemia exhibited a 1.54-fold greater overall risk

of cancer than did the general population. Moreover, patients with

thalassaemia who received blood transfusion had a substantial risk

of haematological malignancy and abdominal cancer. However,

future studies should be conducted to confrm this epiphenomenon

because the fndings were obtained from an observational study.