Contents lists available at ScienceDirect

Clinical Neurology and Neurosurgery

journal homepage: www.elsevier.com/locate/clineuro

Full length article

Trends in the incidence of primary malignant brain tumors in Taiwan andcorrelation with comorbidities: A population-based study

Ya-Jui Lina, Hsiao-Yean Chiub, Meng-Jiun Chiouc, Yin-Cheng Huangd, Kuo-Chen Weid,Chang-Fu Kuoe,f, Jun-Te Hsug, Pin-Yuan Chena,d,⁎

a Department of Neurosurgery, Chang Gung Memorial Hospital, Keelung, Taiwanb School of Nursing, College of Nursing, Taipei Medical University, Taipei, Taiwanc Office for Big Data Research, Chang Gung Memorial Hospital, Taoyuan, Taiwand Department of Neurosurgery, Linkou Chang-Gung Memorial Hospital and School of medicine, Chang Gung University, Taiwane Division of Rheumatology, Allergy and Immunology, Chang Gung Memorial Hospital, Taoyuan, Taiwanf Division of Rheumatology, Orthopaedics and Dermatology, School of Medicine, University of Nottingham, Nottingham, UKg Department of Surgery Chang Gung Memorial Hospital, Taoyuan, Taiwan

A R T I C L E I N F O

Keywords:Primary brain neoplasmEpidemiologyIncidenceSurvivalComorbidity

A B S T R A C T

Objective: Primary malignant brain tumors are relatively uncommon, and their incidence and survival rates haveseldom been reported.Patients and methods: We identified all patients with malignant brain tumors in Taiwan between 1997 and 2012using the National Health Insurance database. We estimated the stratified incidence of malignant brain tumorsby age and sex. We estimated the median 1-, 2-, and 5-year survival, taking comorbidities into account. Trendsfor incidence and survival were analyzed using Joinpoint regression. The incidence in different geographic areaswas also evaluated.Results: A total of 7746 men and 5846 women were identified. The incidence of malignant brain tumor was 3.34(95% CI, 3.09–3.59) per 100,000 person-years in 1997 and 3.82 (95% CI, 3.56–4.08) per 100,000 person-yearsin 2012. The average annual percentage change (APC) of the standardized incidence over this period was 0.1(95% CI, −1.9 to 2.2), suggesting a relatively stable incidence. However, the incidence significantly decreasedbetween 1999 and 2012, with an APC of −1.8 [95% CI, −2.5 to −1.0]. One- and 5-year survival was 53.8%(50.0%–57.5%) and 27.5% (24.1%–30.9%) in 1997 and 67.6% (64.3%–70.7%) and 32.8% (29.6%–35.9%) in2012. The average APC was 1.1 (95% CI, 0.7–1.5) for 1-year survival and 0.2 (95% CI, −1.0–1.4) for 5-yearsurvival. The trend of improvement in the survival rate was seen for short-term but not long-term survival,especially in the group with more comorbidities.Conclusions: A slightly decreased trend in incidence of primary malignant brain tumors was observed inTaiwanese general population since 1999. Over the past 15 years, the short-term survival of malignant braintumors has improved, especially in adults.

1. Introduction

Brain tumors are a mixed group of primary and metastaticneoplasms that exhibit varying degrees of malignancy. Malignantlesions are relatively uncommon, but their incidence has increasedrapidly in highly developed, industrialized countries [1]. Primarymalignant brain tumors have drawn wide attention not only becauseof their poor prognosis, but also for their direct impact on neurologicfunction, psychological health, and quality of life [2]. Over the pastdecades, advances in diagnostic methods have revealed a trend toward

an increased incidence of brain tumors [3,4]. The introduction ofcomputed tomography in the mid-1970s and magnetic resonanceimaging in the mid-1980s improved the detection of these lesions [4].However, recent large-scale cohort- and population-based studies haveproduced mixed results regarding incidence trends. Some studies havefound a level-off or even a slight decrease [1,5,6], while others havefound the opposite [7]. These previous studies focused on primarybrain- or central nervous system tumors; little is known about thetrends in incidence of primary versus metastatic malignant braintumors. The only well-established reported risk factors are ionizing

http://dx.doi.org/10.1016/j.clineuro.2017.05.021Received 17 February 2017; Received in revised form 18 May 2017; Accepted 23 May 2017

⁎ Corresponding author at: Department of Neurosurgery, Chang Gung Memorial Hospital, Keelung, Taiwan.E-mail address: pinyuanc@gmail.com (P.-Y. Chen).

Clinical Neurology and Neurosurgery 159 (2017) 72–82

Available online 25 May 20170303-8467/ © 2017 Published by Elsevier B.V.

MARK

radiation and rare hereditary syndromes such as neurofibromatosis [4].The relation between primary malignant brain tumors and the usage ofcell phones is still a topic of debate [8]. Understanding the epidemiol-ogy is required to facilitate early detection, treatment, and preventionof malignant brain tumors.

Cancer Incidence in Five Continents (CI5) Volume IX and X presentsincidence data from populations all over the world from 1998 to 2002[9] and from 2003 to 2007[10]. This volume covers 11% and 14% ofthe worldwide population respectively. However, Taiwan is not in-cluded in this volume. Therefore, the purpose of our study is to estimatethe trends in incidence over time of primary malignant brain tumorsaccording to sex and age, using the National Health Insurance (NHI)Database, which contains health information for the entire populationof Taiwan.

2. Materials and methods

This study was approved by the Institutional Review Board of ChangGung Memorial Hospital (approval number 201602029B0). As we usedanonymized data, patient consent was exempted. A research grant fromthe Chang Gung Memorial Hospital supported this study.

2.1. Source of data and study population

The primary data source was the NHI database. The Taiwanese NHIwas established in 1995 as a single-payer system requiring all residentsto enroll. In 2012, the coverage rate was exceptional, at 99.5%. TheAdministration of National Health Insurance routinely collected regis-tration- and original claims data generated by the NHI and released theinformation as a single database after anonymization of personaldetails. The database contains comprehensive information regardingbeneficiaries’ personal information, diagnoses, medications, proce-dures, family relationships, vaccines, operations, and fees incurred.Multiple diagnoses recorded in NHI have been validated [9–12], andthe recordings in the database are generally considered accurate.

Patients with primary malignant brain tumors were referred to andtreated by specialists. They were entitled to a waiver for medicalcopayment, granted after a formal review of the relevant clinical andpathological evidence by a commissioned expert committee. In this

study, we identified patients with brain tumors using the registry ofcatastrophic illnesses, which records all patients who received suchbenefits.

2.2. Definition of malignant brain tumor

The primary definition of malignant brain tumor used for this studywas a physician-recorded primary diagnosis of malignant brain tumor(International Classification of Diseases, Ninth Revision [ICD-9]code:191) recorded in the catastrophic illness registry. We obtaineddata from all patients in the NHI database with a primary malignantbrain tumor between 1997 and 2012. To ascertain the validity of braincancer diagnoses in the NHI database, we compared the diagnoses ofprimary malignant brain tumor in the NHI with the National CancerRegistry. Agreement between the 2 databases was good, with asensitivity of 0.86 and a specificity of 1.

2.3. Estimation of prevalence and incidence

Incident patients were those who had no evidence of disease prior tothe first of January of each calendar year but who developed amalignant brain tumor during that year. To be eligible, beneficiarieshad to have at least 1 year of registration in the database prior to thefirst of January of each calendar year. We constructed at-risk cohortsthat included all individuals registered during a given calendar yearwho had no history of a malignant brain tumor diagnosis before the firstof January of that year. The incidence was calculated using the numberof patients with incident primary malignant brain tumors during acalendar year as the numerator, and the total person-years at-riskpopulation of that same year as the denominator.

2.4. Standardized incidence of malignant brain tumor

To determine trends in the incidence of primary malignant braintumor, we calculated age- and sex-standardized incidence using thepopulation structure in Taiwan in 2012 as the reference.

Table 1Clinical characteristics.

Male (n = 7746) Female (n = 5846) P-value

Age (years), mean ± standard deviation 49.38 ± 22.25 49.79 ± 21.76 0.28360–17 1076 (13.89) 718 (12.28) 0.00418–64 4344 (56.08) 3424 (58.57)≥65 2326 (30.03) 1704 (29.15)

Place of residence, n (%) 0.0351Urban 4322 (55.80) 3395 (58.07)Suburban 2520 (32.53) 1770 (30.28)Rural 775 (10.01) 589 (10.08)Unknown 129 (1.67) 92 (1.57)

Income level, n (%) < 0.0001Quintile 1 1602 (20.68) 1076 (18.41)Quintile 2 853 (11.01) 775 (13.26)Quintile 3 2291 (29.58) 1822 (31.17)Quintile 4 1296 (16.73) 1065 (18.22)Quintile 5 1615 (20.85) 1074 (18.37)Unknown 89 (1.15) 34 (0.58)

Occupation, n (%) < 0.0001Dependent of the insured 2394 (30.91) 2336 (39.96)Civil servant, teacher, military personnel, veteran 476 (6.15) 173 (2.96)Nonmanual worker, professional 1416 (18.28) 926 (15.84)Manual worker 2445 (31.56) 1959 (33.51)Other 935 (12.07) 424 (7.25)Unknown 80 (1.03) 28 (0.48)

Charlson index (mean ± standard deviation) 4.32 ± 2.91 4.23 ± 2.85 0.0861

Y.-J. Lin et al. Clinical Neurology and Neurosurgery 159 (2017) 72–82

73

Fig. 1. Differences, by sex, in trends for the standardized incidence of brain tumor in overall (a), age<20 years (b), and age ≥ 20 years (c) in Taiwan between 1997 and 2012.AAPC: Average annual percentage change.

Y.-J. Lin et al. Clinical Neurology and Neurosurgery 159 (2017) 72–82

74

2.5. Mortality and causes of death

Survival status and causes of death between 1997 and 2012 wereascertained using the National Death Registry in Taiwan, which recordsthe deaths of all citizens, coding the causes of death from deathcertificates. The accuracy of the coding has been validated by previousstudies [13]. We estimated 1-year survival rates for calendar years

1997–2012, 2-year survival rates for 1997–2011, and 5-year survivalrates for 1997–2008.

2.6. Statistical analysis

The 95% confidence interval (CI) for incidence was derived basedon the assumption of a Poisson distribution for the observed number of

Table 2Standardized incidence of brain tumors (1997–2012).

Year Total Male Female

Patients (n) Standardized Patients (n) Standardized Patients (n) Standardized

1997 676 3.85 (3.54–4.16) 409 4.59 (4.12–5.07) 267 3.13 (2.73–3.53)1998 767 4.34 (4.02–4.67) 428 4.65 (4.18–5.12) 339 4.05 (3.59–4.50)1999 891 5.02 (4.67–5.36) 506 5.52 (5.01–6.03) 385 4.52 (4.05–4.99)2000 874 4.81 (4.47–5.14) 503 5.38 (4.88–5.87) 371 4.25 (3.80–4.71)2001 897 4.87 (4.53–5.20) 503 5.24 (4.77–5.72) 394 4.50 (4.04–4.96)2002 829 4.41 (4.10–4.72) 448 4.59 (4.14–5.03) 381 4.24 (3.80–4.68)2003 896 4.71 (4.39–5.03) 517 5.30 (4.83–5.78) 379 4.13 (3.70–4.56)2004 897 4.61 (4.30–4.92) 512 5.18 (4.72–5.65) 385 4.05 (3.63–4.47)2005 889 4.56 (4.25–4.87) 501 5.11 (4.65–5.57) 388 4.03 (3.61–4.44)2006 880 4.39 (4.09–4.69) 486 4.80 (4.36–5.24) 394 3.99 (3.59–4.39)2007 792 3.93 (3.65–4.20) 444 4.36 (3.94–4.77) 348 3.51 (3.13–3.88)2008 853 4.13 (3.84–4.41) 504 4.86 (4.43–5.29) 349 3.41 (3.05–3.77)2009 824 3.94 (3.67–4.21) 480 4.60 (4.18–5.01) 344 3.29 (2.94–3.64)2010 927 4.39 (4.11–4.67) 542 5.15 (4.71–5.58) 385 3.65 (3.28–4.02)2011 875 4.10 (3.83–4.37) 490 4.64 (4.23–5.05) 385 3.58 (3.22–3.93)2012 825 3.82 (3.56–4.08) 473 4.44 (4.04–4.84) 352 3.22 (2.88–3.56)

Table 3Joinpoint analysis of brain tumor incidence by age and sex in Taiwan (1997–2012).

Incidence Trend 1 Trend 2

(per 100,000 person-years)

1997 2012 Average APC Years APC (95%CI) Years APC (95%CI)

OverallTotal 3.85 (3.54–4.16) 3.82 (3.56–4.08) 0.1 (−1.9 to 2.2) 1997–1999 13.4 (−3.7 to 33.4) 1999–2012 −1.8 (−2.5 to −1.0)*

Male 4.59 (4.12–5.07) 4.44 (4.04–4.84) 0.1 (−2.4 to 2.8) 1997–1999 9.1 (−11.4 to 34.3) 1999–2012 −1.2 (−2.1 to −0.2)*

Female 2.68 (2.36–3.00) 3.22 (2.88–3.56) 0.1 (−2.0 to 2.3) 1997–1999 18.8 (−0.3 to 41.5) 1999–2012 −2.5 (-3.3 to −1.7)

Age < 20 yTotal 2.05 (1.69–2.41) 2.35 (1.90–2.80) 0.8 (−0.2 to 1.7) N/A N/AMale 2.32 (1.79–2.86) 2.75 (2.07–3.43) 1.1 (−0.3 to 2.5) N/A N/AFemale 1.75 (1.28–2.22) 1.91 (1.32–2.50) 0.4 (−1.3 to 2.0) N/A N/A

Age ≥ 20 yTotal 4.31 (3.93–4.69) 4.2 (3.89–4.50) 0.1 (−2.1 to 2.2) 1997–1999 14.8 (−3.4 to 36.3) 1999–2012 −2.0 (−2.8 to −1.3)*

Male 5.21 (4.62–5.80) 4.9 (4.43–5.38) 0.1 (−2.7 to 2.8) 1997–1999 10.3 (−11.5 to 37.7) 1999–2012 −1.5 (-2.4 to −0.4)Female 3.46 (2.98–3.94) 3.53 (3.14–3.92) 0.1 (−2.5 to 2.6) 1997–1999 20.3 (−2.0 to 47.7) 1999–2012 −2.8 (−3.6 to −1.9)*

* P < 0.05.

Fig. 2. Differences, by sex and age, in trends for the incidence of brain tumor in Taiwan in 2012.

Y.-J. Lin et al. Clinical Neurology and Neurosurgery 159 (2017) 72–82

75

Table4

Joinpo

intan

alysis

ofsurvival

bysexin

Taiw

an(199

7–20

12).

Ove

rall

Age

<20

yAge

≥20

y

Survival

Survival

Survival

1997

EndYear

Ave

rage

APC

1997

EndYear

Ave

rage

APC

1997

EndYear

Ave

rage

APC

Ove

rall

Total

1-ye

arsurvival

53.8

(50.0–

57.5)

67.6

(64.3–

70.7)

1.1

(0.7–1

.5)*

67.2

(58.3–

74.6)

81.5

(72.5–

87.8)

0.4

(−0.1to

1.0)

50.7

(46.5–

54.8)

65.7

(62.1–

69.0)

1.3

(0.9–1

.7)*

2-ye

arsurvival

38.4

(34.7–

42.1)

46.6

(43.2–

49.8)

0.9

(0.3–1

.6)*

64.1

(55.1–

71.7)

73.2

(63.0–

81.0)

0.2

(-0.5to

0.9)

32.4

(28.5–

36.3)

43.4

(39.9–

46.8)

1.4

(0.6–2

.2)*

5-ye

arsurvival

27.5

(24.1–

30.9)

32.8

(29.6–

35.9)

0.2

(−1.0to

1.4)

52.9

(43.8–

61.1)

64.5

(54.8–

72.6)

0.3

(-1.1to

1.6)

21.5

(18.1–

25.1)

28.1

(24.9–

31.3)

0.6

(-0.9to

2.1)

Male

1-ye

arsurvival

51.8

(46.9–

56.6)

65.7

(61.3–

69.8)

1.2

(0.7–1

.7)*

68.9

(57.0–

78.1)

85.6

(74.2–

92.2)

0.6

(0.1–1

.2)*

48.1

(42.6–

53.3)

62.7

(57.8–

67.2)

1.4

(0.9–2

.0)*

2-ye

arsurvival

35.2

(30.6–

39.8)

44.6

(40.1–

48.9)

1.1

(0.2–2

.0)*

66.2

(54.2–

75.7)

79.8

(67.1–

88.0)

0.4

(−0.5to

1.2)

28.3

(23.6–

33.2)

39.7

(35.1–

44.3)

1.6

(0.4–2

.7)*

5-ye

arsurvival

24.9

(20.8–

29.2)

29.6

(25.6–

33.6)

0.1

(−1.6to

1.9)

53.7

(41.7–

64.3)

69.0

(56.2–

78.8)

0.4

(−2.0to

2.8)

18.5

(14.6–

22.9)

23.7

(19.8–

27.8)

0.7

(-1.4to

2.8)

Female

1-ye

arsurvival

56.9

(50.8–

62.6)

70.2

(65.1–

74.7)

0.9

(0.5–1

.3)*

64.8

(50.5–

75.9)

75.0

(58.5–

85.7)

0.1

(−0.8to

1.0)

54.9

(48.0–

61.3)

69.6

(64.1–

74.4)

1.1

(0.7–1

.6)*

2-ye

arsurvival

43.4

(37.3–

49.2)

49.1

(44.0–

54.0)

0.8

(0.2–1

.3)*

61.1

(46.8–

72.6)

61.8

(43.4–

75.7)

−0.1

(−1.7to

1.5)

38.8

(32.3–

45.3)

47.9

(42.5–

53.0)

1.2

(0.6–1

.7)*

5-ye

arsurvival

31.4

(25.8–

37.0)

37.4

(32.3–

42.4)

0.3

(−1.2to

1.8)

51.7

(37.6–

64.0)

58.0

(42.3–

70.8)

0.3

(−2.0to

2.6)

26.2

(20.4–

32.3)

34.3

(29.0–

39.7)

0.5

(−1.1to

2.3)

CCI<

31-ye

arsurvival

64.9

(59.6–

69.6)

79.6

(74.3–

83.9)

0.8

(0.2–1

.4)*

68.7

(57.5–

77.5)

92.3

(78.0–

97.5)

0.6

(−0.3to

1.6)

63.7

(57.6–

69.1)

77.4

(71.5–

82.3)

1.0

(0.3–1

.6)*

2-ye

arsurvival

48.2

(42.8–

53.3)

61.5

(55.5–

67.0)

0.7

(−0.2to

1.6)

63.9

(52.5–

73.1)

79.8

(67.1–

88.0)

0.9

(−0.2to

2.0)

43.3

(37.3–

49.1)

57.7

(51.0–

63.7)

0.9

(−0.2to

1.9)

5-ye

arsurvival

35.7

(30.7–

40.8)

45.2

(39.4–

50.9)

1(−

0.6to

2.5)

55.3

(43.9–

65.2)

73.4

(57.2–

84.3)

1.1

(−0.4to

2.6)

29.6

(24.2–

35.2)

40.5

(34.3–

46.6)

1.2

(−0.8to

3.4)

CCI≥

31-ye

arsurvival

42.0

(36.6–

47.3)

57.4

(53.3–

61.2)

1.6

(1.0–2

.3)*

64.4

(48.7–

76.5)

76.1

(62.5–

85.4)

0.8

(0.01–

1.5)

*38

.4(32.7–

44.1)

55.5

(51.2–

59.5)

1.8

(1.1–2

.6)*

2-ye

arsurvival

27.9

(23.2–

32.9)

39.8

(35.8–

43.7)

1.9

(1.0–2

.7)*

64.4

(48.7–

76.5)

65(50.8–

76.1)

−0.2

(−1.3to

0.8)

22.1

(17.4–

27.1)

37.2

(33.2–

41.3)

2.4

(1.2–3

.6)*

5-ye

arsurvival

18.6

(14.6–

23.0)

26.7

(22.9–

30.8)

1.0

(−1.0to

3.0)

55.3

(43.9–

65.2)

59.1

(46.5–

69.7)

−0.3

(−2.1to

1.5)

13.9

(10.1–

18.2)

22(18.5–

25.8)

1.3

(−1.4to

4.1)

*P

<0.05

.

Y.-J. Lin et al. Clinical Neurology and Neurosurgery 159 (2017) 72–82

76

incident patients. We used the Joinpoint Regression Program (version4.0.4) to estimate trends in the incidence and survival rates ofmalignant brain tumors. The program used the Bayesian informationcriterion to generate different numbers of “joinpoints” that partition thelinear trend into periods with statistically different slopes using thebest-fit data series [14]. A maximum of 2 joinpoints were used todetermine statistical significance for the trend. Annual percentagechanges (APCs) for each segment were calculated. The secular trendsfor incidence and survival were calculated for patients with a Charlsoncomorbidity score of 0–2 and ≥ 3. The Charlson comorbidity index(CCI) was used to evaluate patients’ burden of comorbidities and toestimate the risk of death from 17 diagnostic conditions (weightedscore 1: myocardial infarction, congestive heart disease, peripheralvascular disease, cerebrovascular disease, dementia, chronic pulmonarydisease, connective tissue disease, peptic ulcer disease, mild liverdisease, diabetes without complications; score 2: diabetes with compli-cations, hemiplegia and paraplegia, renal disease, any malignancyincluding leukemia and lymphoma without metastasis; score 3: mod-erate or severe liver disease; score 6: malignant metastasis and acquiredimmune deficiency syndrome). The significance level was set at 0.05.All statistical analyses were performed using SAS statistical software,version 9.3.

3. Results

We identified 7746 male and 5846 female patients with primarymalignant brain tumors in Taiwan between 1997 and 2012. The meanage at diagnosis was 49 years. Patient characteristics are shown inTable 1.

3.1. Incidence of primary malignant brain tumor between 1997 and 2012

Fig. 1 shows the temporal trends in the standardized incidence ofmalignant brain tumors between 1997 and 2012. The APC of thestandardized incidence was 0.1 (95% CI, −1.9 to 2.2). This was notstatistically significant, suggesting that the overall incidence of primary

malignant brain tumor was essentially the same throughout the studyperiod, taking into account age and sex (Tables 2 and 3). However,Joinpoint analysis indicated that the incidence decreased between 1999and 2012, with an APC of −1.8 [95% CI, −2.5 to −1.0]. This trendwas particularly evident in adults,with an APC of −2.0 [95% CI, −2.8to −1.3].

3.2. Incidence of primary malignant brain tumor in 2012

There were a total 21,582,604 person-years of follow-up for 2012. Atotal of 825 incident cases of primary malignant brain tumor wereidentified, with an overall incidence of 3.82 per 100,000 person-years[95% CI, 3.56–4.08]. The incidence was 4.44 [95% CI, 4.04–4.84] inmen and 3.22 [95% CI, 2.88–3.56] in women. Overall, the incidencewas higher in the adult group, at 4.2 per 100,000 person-years [95% CI,3.89–4.5]. In the pediatric group, the incidence was 2.35 [95% CI,1.9–2.8] per 100,000 person-years (Table 3). Men had a higherincidence than women at all ages (Fig. 2). The incidence of primarymalignant brain tumor in both men and women was low in thoseyounger than 40 years of age, but thereafter increased rapidly until theage of 70 years.

3.3. Survival rates of primary malignant brain tumors between 1997 and2012

Of the 13592 patients with primary malignant brain tumorsdiagnosed between 1997 and 2012, the overall 1-, 2-, and 5-yearsurvival was 65.7% (95% CI, 62.1–69.0), 43.4% (95% CI, 39.9–46.8),and 28.1% (95% CI, 24.9–31.3). Overall, the 1-year survival rate was53.8% (95% CI, 50.0–57.5) in 1997 and 67.6% (64.3%–70.7%) in 2012(Table 4). The APC was 1.1 (0.7–1.5) for this period, suggesting animprovement in short-term mortality; and the average APC of 2-yearsurvival between 1997 and 2011 was 0.9 (0.3–1.6). However, theaverage APC of 5-year survival was 0.2 (−1.0 to 1.4), suggesting thatlong-term survival did not improve (Fig. 3). Similar trends wereobserved in overall survival in male and female groups, those with a

Fig. 3. Trends for 1-, 2-, and 5-year survival for brain tumor, according to age and sex, in Taiwan between 1997 and 2012.AAPC: Average annual percentage change.(1a) Male, (1b) Female, (2a) Male, age < 20 years, (2b) Female, age< 20 years, (3a) Male, age ≥20 years, (3b) Female, age ≥ 20 years.

Y.-J. Lin et al. Clinical Neurology and Neurosurgery 159 (2017) 72–82

77

CCI > 3, and in the adult group. In the pediatric group, only 1-yearsurvival in males (APC, 0.6 [95% CI, 0.1–1.2]) and 1-year survival inthose with a CCI > 3 (APC, 0.8 [95% CI, 0.01–1.5]) improved (Fig. 4).

3.4. Geographic distribution of primary malignant brain tumors in Taiwan

Fig. 5 shows the geographic distribution of incidence in Taiwan in2012. The higher incidence areas were Miaoli, Tainan, and Hualiencounties. For patients below 20 years of age, Hsinchu county, Taichungcity, and Taitung county were high-incidence areas. For patients above20 years of age, Miaoli, Hualien, and Nantou counties were high-incidence areas.

4. Discussion

Cancer is the leading cause of death worldwide, with 8.2 millionpeople dying each year. In Taiwan, cancer was also the leading cause ofdeath in 2014 and 2015 (http://iiqsw.mohw.gov.tw/st3/index.html).Although the incidence of primary malignant brain tumors is relativelylow compared with other cancers, the former usually cause severeneurologic dysfunction and influence quality of life. The trends ofincidence and survival for primary malignant brain tumors, using thedata of the Taiwanese NHI, give us a chance to understand the changes

in incidence and survival in this area of Asia.

4.1. Changes in incidence of primary malignant brain tumor

The worldwide annual age-standardized incidence of primarymalignant brain tumors is 3.7 per 100,000 for men and 2.6 per100,000 for women [8,10]. Rates appear to be higher in moredeveloped countries (men, 5.8 per 100,000; women, 4.1 per 100,000)than in less developed countries (men, 3.0 per 100,000; women, 2.1 per100,000) [12]. In the United States, the annual age-adjusted incidenceof primary malignant brain tumor from 2008 to 2012 was 7.44, 7.35,7.21, 7.12, and 7.02 per 100,000 [1]. In comparison, the annual age-adjusted incidence (age-adjusted to the United States standard popula-tion in 2000) In Taiwan from 2008 to 2012 was 4.06, 3.91, 4.34, 4.05,and 3.85 per 100,000. A relatively low incidence in Taiwan was found,considering that it is a developed country. One may expect an increasedincidence due to early detection, as with other cancers, with improve-ments in global public health. However, there are few reports describ-ing trends in the incidence of primary malignant brain tumors inTaiwan from 1997 to 2012.

Inconsistent with the existing literature comparing the incidence ofprimary brain tumors between the United States and Taiwan from 2002to 2010 using the databases of the Unites States and Taiwan Cancer

Fig. 3. (continued)

Y.-J. Lin et al. Clinical Neurology and Neurosurgery 159 (2017) 72–82

78

Registry [15], we found that primary malignant brain tumors seem tobe slowly decreasing from 1999 to 2012; this is compatible withfindings in the United States [1], where the incidence of primarymalignant brain tumors gradually decreased between 2008 and 2012.Differences in databases (i.e., National Health Insurance versus CancerRegistry) or the codes use (i.e., ICD-9 CM versus InternationalClassification for Disease, Oncology, third edition) might be reasonsfor the discrepancy in results between this study and the previousreport [15].

Of note, the findings from joinpoint analysis suggested that theincidence decreased between 1999 and 2012, with an APC of −1.8[95% CI, −2.5 to −1.0]. Several factors, including screening practices,disparities in lifestyle, environmental exposure, and health policy,might explain this matter [15]. However, this study could not ascertainthe association due to the limitation of study design.

In the pediatric group, primary malignant brain tumors are the mostcommon cancer among those aged 0–19 years, ranging from 2.27 to4.81 per 100,000 population, with an average annual age-adjustedincidence of 5.57 per 100,000 in the United States [1]. In Kuwait, theoverall incidence between 1995 and 2011 was 1.12 per 100,000population. In Japan, the overall incidence of primary malignant braintumor in ages 0–14 years is 3.61 per 100,000; in Denmark, Finland,Norway, and Sweden, the incidence is 4.20 per 100,000 persons [16].Our data indicate that the incidence of primary malignant brain tumorranges from 1.70 to 2.35 per 100,000 between 1998 and 2012. There isno increasing or decreasing trend in the pediatric age group. This is arelatively low incidence compared with developed countries (e.g.,United states, Japan, Northern Europe) [16].

The incidence of all brain- and central nervous system tumors ishighest in the group aged 85 years and over (83.14 per 100,000) andlowest among children and adolescents aged 0–19 years (5.57 per100,000) in the United States [1], where approximately 55.0% occur in

males and 45% in females [13]. In Taiwan, a relatively low incidencewas noted in patients aged less than 20 years, while an increasedincidence was found in those aged over 40 years. The highest incidenceis in those aged over 80 years. Men have higher rates of primarymalignant brain tumors, especially those older than 40 years of age. Theincidence in Taiwan and the United States is comparable.

Cancer Incidence in Five Continents (CI5) Volume IX and X showsregional variations in the incidence of malignant brain tumors (age-standardized incidence adjusted to Segi-Doll world population), with ahigher incidence seen in Western, developed countries. Asian countriestend to display lower incidence. In Japan, the age-standardizedincidence in males is 2.1–3.5 per 100,000 and in females is 1.5–2.5per 100,000. In Korea, the male age-standardized incidence is 2.5–3.6and in females is 2.1–3.2. In Singapore, the male age-standardizedincidence is 2.0–3.2 and in females is 1.2–2.4 [9,10]. In our Taiwanesedata, age-standardized incidence (age-adjusted to Segi-Doll worldpopulation) showed 4.94–6.09 in male and 3.56–4.78 in female from1998 to 2007. We observed a relatively high age-standardized inci-dence relative to other Asian countries. However, males have a higherincidence of primary malignant brain tumor in most countries, includ-ing Taiwan.

4.2. Improvement in survival

Overall survival rate of primary malignant brain tumor in Taiwanfrom another publication showed 1-year, 2-year, and 5-year survivalrate are 68.1%, 49.9%, and 34.3% [15]. These data are similar to ours,but higher than that in Unites States. The reasons may be largerproportion of glioblastoma multiforme and lower insurance coveragerate [15].

Primary malignant brain tumors are the most common solid tumorand the second leading cause of cancer-related death in individuals

Fig. 4. Trends for 1-, 2-, and 5-year survival for brain tumor, according to Charlson index, in Taiwan between 1997 and 2012.AAPC: Average annual percentage change.(1a) Charlson index<3, all patients, (1b) Charlson index ≥3, all patients, (2a) Charlson index< 3, age<20 years, (2b) Charlson index ≥ 3, age<20 years, (3a) Charlson index< 3,age ≥20 years, (3b) Charlson index ≥3, age ≥20 years.

Y.-J. Lin et al. Clinical Neurology and Neurosurgery 159 (2017) 72–82

79

aged 0–19 years in the United States and Canada [16]. The overall 1-year and 5-year survival rates are 85.2% and 72.3% for pediatric braintumors in the United States. Our Taiwanese data show a 1-year and 5-year overall survival rate of 81.5% and 64.5%, lower than the rates inthe United States. In addition, the 1-year survival rate in males (APC,0.6 [95% CI, 0.1–1.2]) and the 1-year survival in individuals with aCCI > 3 (APC, 0.8 [95% CI, 0.01–1.5]) improved from 1998 to 2012.This means that medical progress in Taiwan improved the short-termsurvival of pediatric brain tumors in patients with comorbidities, butthere was no effect on long-term survival.

In adults, we observed a significant increase in overall, male, andfemale 1-year and 2-year survival. However, the 5-year survival in thesesame groups did not improve significantly. In other words, short-termsurvival improved, but long-term survival did not. In the United States,the 5- and 10-year survival rates for primary malignant brain tumorsare 29.1% and 25.3% according to data from the American CancerSociety website (www.cancer.org accessed on March 1, 2008) [12].These findings are compatible with our data. Survival times areincreasing due to innovations in diagnosis and therapy [11]. Forexample, chemotherapy using temozolomide with concurrent radio-therapy has become a standard treatment for glioblastoma, whichaccounts for 54% of primary malignant brain tumors since 2005 in theUnited States [11] and in Taiwan. This may be among the reasons for

the improved short-term survival seen over recent years.

4.3. Comorbidities influences survival

Several investigators have attempted to determine whether there isa relation between comorbidity and cancer prognosis. Comorbiditiesmay modify the treatment of older cancer patients. Numerous reports,notably epidemiologic studies, have observed that comorbidities affectboth treatment and prognosis, making it difficult to separate therespective contributions of the comorbidity, functional status, andtreatment reduction to the prognosis [17]. Research also shows thatpreexisting diabetes mellitus and an elevated body mass index areindependent risk factors for poor outcome in patients with high-gradeglioma [19,20]. The CCI takes 17 conditions into consideration andpredicts 10-year mortality. The CCI significantly impacts patient out-comes and may be employed for preoperative patient risk stratification[18].

We noted that overall survival improved for both 1-year and 2-yeardata. Taking comorbidities into consideration, the improvement in thegroup with a CCI < 3 is only seen in 1-year survival for both theoverall group and the adult group, but not the pediatric group.Improvement in 2-year survival was noted in the adult group but notin the pediatric group. Several studies have reported that comorbidities

Fig. 4. (continued)

Y.-J. Lin et al. Clinical Neurology and Neurosurgery 159 (2017) 72–82

80

may worsen the prognosis; therefore, if comorbid disease is underbetter control, the prognosis will be better. This may be the cause of theimproved survival seen in the group with a CCI > 3 between 1998 and2012.

We may be the first to report that controlling comorbidities couldimprove short-term survival for all primary malignant brain tumors, notjust glioblastomas. Our data indicate that controlling comorbidities hasan impact on survival in malignant brain tumors, especially in adults.They also imply that further improvement in long-term survival mayrequire more effective anticancer treatment.

4.4. Association between geographic variations in malignant brain tumor

Our findings suggested that some areas yielded higher incidencerates of primary malignant brain tumor than other counties. Note thatthese counties were not all high population density areas, indicating nosubstantial association between population density and the incidence ofprimary malignant brain tumors. Several risk factors have beenrecognized to be associated with primary malignant brain tumors, suchas chemical exposure, radiation exposure, and certain genetic syn-dromes [15,21]. However, we could not determine the mechanismunderlying this association. Therefore, the nature of the relationshipbetween must be further explored.

4.5. Limitations

This is a retrospective study based on the NHI database. We werenot able to differentiate between various types of intracranial pathol-ogy. Different pathology may occur in different age groups and sexesand may receive variable treatment, therefore contributing a differentpercentage of improvement for survival.

4.6. Conclusions

A slightly decreased trend in incidence of primary malignant braintumors was observed in Taiwanese general population since 1999. Overthe past 15 years, we have seen improvement in the short-term survivalof primary malignant brain tumors, especially in adults.

Author contributions

Study concept and design: Pin-Yuan Chen, Chang-Fu Kuo;Acquisition of data: Hsiao-Yean Chiu; Analysis and interpretation ofdata: Meng-Jiun Chiou; Drafting of the manuscript: Ya-Jui Lin; Criticalrevision of the manuscript for important intellectual content: allauthors; Statistical analysis: Meng-Jiun Chiou; Funding acquisition:Chang-Fu Kuo; Administrative, technical, and material support: Kuo-

Fig. 5. Geographic variation in the incidence of brain tumor in Taiwan in 2012.(a) All patients, (b) Age<20 years, (c) Age ≥20 years.

Y.-J. Lin et al. Clinical Neurology and Neurosurgery 159 (2017) 72–82

81

Chen Wei; Study supervision: Chang-Fu Kuo.All authors had full access to all of the data in the study and take

responsibility for the integrity of the data and the accuracy of the dataanalysis.

Sources of funding and support

This work was funded by CORP3E0152.

Role of the sponsors

The sponsors of the study, the Chang Gung Memorial Hospital, theNational Science Council, and the University of Nottingham, had norole in the design and conduct of the study; the collection, management,analysis, and interpretation of the data, or the preparation, review,approval, or decision to submit the manuscript for publication.

Disclaimer

This study is based in part on date from the National HealthInsurance Research Database provided by the Administration ofNational Health Insurance, Ministry of Health and Welfare, andmanaged by the National Health Research Institutes. The interpretationand conclusions contained herein do not represent the position of theAdministration of National Health Insurance or the National HealthResearch Institutes.

Acknowledgments

The authors wish to acknowledge the support of the Maintenanceproject for Center for Big Data Analytics and Statistics (GrantCLRPG3D0043) from Chang Gung Memorial Hospital (https://www.cgmh.org.tw/) for study design and monitoring, analysis and inter-pretation of data.

References

[1] Q.T. Ostrom, H. Gittleman, J. Fulop, et al., CBTRUS statistical report: primary brainand central nervous system tumors diagnosed in the United States in 2008–2012,

Neuro Oncol. 17 (Suppl. 4) (2015) iv1–iv62.[2] R. Liu, M. Page, K. Solheim, S. Fox, S.M. Chang, Quality of life in adults with brain

tumors: current knowledge and future directions, Neuro Oncol. 11 (2009) 330–339.[3] S. Hoffman, J.M. Propp, B.J. McCarthy, Temporal trends in incidence of primary

brain tumors in the United States, 1985–1999, Neuro Oncol. 8 (2006) 27–37.[4] I. Deltour, C. Johansen, A. Auvinen, M. Feychting, L. Klaeboe, J. Schuz, Time trends

in brain tumor incidence rates in Denmark, Finland, Norway, and Sweden,1974–2003, J. Natl. Cancer Inst. 101 (2009) 1721–1724.

[5] V.K. Ho, J.C. Reijneveld, R.H. Enting, et al., Changing incidence and improvedsurvival of gliomas, Eur. J. Cancer 50 (2014) 2309–2318.

[6] E. Nomura, A. Ioka, H. Tsukuma, Trends in the incidence of primary intracranialtumors in Osaka, Japan, Jpn. J. Clin. Oncol. 41 (2011) 291–294.

[7] N. Igissinov, S. Akshulakov, T. Kerimbayev, et al., Malignant tumors of the centralnervous system in Kazakhstan: component analysis of incidence dynamics, AsianPac. J. Cancer Prev. 16 (2015) 2289–2293.

[8] V.G. Khurana, C. Teo, M. Kundi, L. Hardell, M. Carlberg, Cell phones and braintumors: a review including the long-term epidemiologic data, Surg. Neurol. 72(2009) 205–214.

[9] M.P. Curado, B. Edwards, H.R. Shin, Cancer Incidence in Five Continents vol. IX,International Agency for Research on Cancer, Lyon, France, 2007.

[10] D. Forman, F. Bray, D.H. Brewster, Cancer Incidence in Five Continents vol. X,International Agency for Research on Cancer, Lyon, France, 2014.

[11] A. Wöhrer, Epidemiology and brain tumours: practical usefulness, Eur. Assoc.Neuro Oncol. Mag. 3 (2013) 56–60.

[12] M.L. Bondy, M.E. Scheurer, B. Malmer, et al., Brain tumor epidemiology: consensusfrom the brain tumor epidemiology consortium, Cancer 113 (Suppl. 7) (2008)1953–1968.

[13] T.H. Lu, W.Y. Shau, T.P. Shih, Factors associated with errors in death certificatecompletion. A national study in Taiwan, J. Clin. Epidemiol. 54 (3) (2001) 232–238.

[14] H.J. Kim, M.P. Fay, E.J. Feuer, D.N. Midthune, Permutation tests for joinpointregression with applications to cancer rates, Stat. Med. 19 (2000) 335–351.

[15] Li-Nien Chien, Haley Gittleman, Quinn T. Ostrom, Comparative brain and centralnervous system tumor incidence and survival between the United States and Taiwanbased on population-based registry, Front. Public Health 4 (2016) 151.

[16] K.J. Johnson, J. Cullen, J.S. Barnholtz-Sloan, et al., Childhood brain tumorepidemiology: a brain tumor epidemiology consortium review, Cancer Epidemiol.Biomark. Prev. 23 (2014) 2716–2736.

[17] M. Extermann, Interaction between comorbidity and cancer, Cancer Control 14(2007) 13–22.

[18] G. Ening, F. Osterheld, D. Capper, K. Schmieder, C. Brenke, Charlson comorbidityindex: an additional prognostic parameter for preoperative glioblastoma patientstratification, J. Cancer Res. Clin. Oncol. 141 (2015) 1131–1137.

[19] J.L. Fisher, S. Palmisano, J.A. Schwartzbaum, T. Svensson, S. Lönn, Comorbidconditions associated with glioblastoma, J. Neurooncol. 116 (2014) 585–591.

[20] L.B. Chambless, S.L. Parker, L. Hassam-Malani, M.J. McGirt, R.C. Thompson, Type 2diabetes mellitus and obesity are independent risk factors for poor outcome inpatients with high-grade glioma, J. Neurooncol. 106 (2012) 383–389.

[21] S.M. Chang, I.F. Parney, W. Huang, Patterns of care for adults with newly diagnosedmalignant glioma, JAMA 293 (5) (2005) 557–564.

Y.-J. Lin et al. Clinical Neurology and Neurosurgery 159 (2017) 72–82

82