Iodine deficiency and thyroid cancer trends in three regions of Thailand, 1990–2009

Cancer Epidemiology 43 (2016) 92–99

Contents lists available at ScienceDirect

Cancer Epidemiology The International Journal of Cancer Epidemiology, Detection, and Prevention journal homepage: www.cancerepidemiology.net

Iodine deficiency and thyroid cancer trends in three regions of Thailand, 1990–2009 Susanna D. Mitroa,1, Laura S. Rozekb , Patravoot Vatanasaptc , Krittika Suwanrungruangd , Imjai Chitapanaruxe , Songpol Srisukhof , Hutcha Sriplungg,** , Rafael Mezaa,* a

Department of Epidemiology, University of Michigan School of Public Health, Ann Arbor, MI, United States Department of Environmental Health Sciences, University of Michigan School of Public Health, Ann Arbor, MI, United States Department of Otorhinolaryngology, Faculty of Medicine, Khon Kaen University, Thailand d Cancer Unit, Srinagarind Hospital, Faculty of Medicine, Khon Kaen University, Thailand e Division of Therapeutic Radiology and Oncology, Department of Radiology, Faculty of Medicine, Chiang Mai University, Thailand f Division of Head, Neck and Breast Surgery, Department of Surgery, Faculty of Medicine, Chiang Mai University, Thailand g Epidemiology Unit, Faculty of Medicine, Prince of Songkla University, Songkhla, Thailand b c

A R T I C L E I N F O

A B S T R A C T

Article history: Received 27 March 2016 Received in revised form 16 June 2016 Accepted 4 July 2016 Available online xxx

Background: Iodine deficiency may play a role in thyroid cancer carcinogenesis. Because Thailand has region-specific historical iodine deficiency, it is ideal to evaluate the potential impact of recent national iodine supplementation policies on thyroid cancer incidence trends. Methods: We examined thyroid cancer trends in Thailand from 1990 to 2009 in three geographically separated populations (Songkhla Province [south], Chiang Mai Province [north], and Khon Kaen Province [northeast]), each with a different historical prevalence of iodine deficiency. We used Joinpoint analysis and age-period-cohort (APC) models to investigate trends in thyroid cancer incidence. Results: Pooled incidence of papillary cancers significantly increased (Males APC: 2.0, p < 0.05; Females APC: 7.3 [1990–2001, p < 0.05], 2.1 [2001–2009]) and incidence of follicular cancers significantly decreased (Males APC: 5.2, p < 0.05; Females APC: 4.3 [1990–1998, p < 0.05], 12.3 [1998–2001], 17.0 [2001–2005, p < 0.05], 8.2 [2005–2009]) in both males and females between 1990 and 2009. The largest increases in papillary cancer incidence, and the largest decreases in follicular cancer incidence, occurred in historically iodine-deficient regions. Interestingly, the significant histological changes coincided with Thailand’s most recent national iodination policy. The thyroid cancer trends in females were better explained by period effects than cohort effects. Conclusions: This study adds to the research indicating that papillary carcinoma incidence increases, and follicular carcinoma incidence decreases, as population-level iodine deficiency declines, and suggests that iodine exposure may affect late stages of thyroid carcinogenesis. However, our findings are limited by the ecological study design and lack of data prior to iodine supplementation. ã 2016 Elsevier Ltd. All rights reserved.

Keywords: Thyroid neoplasms Iodine Epidemiology Thailand

1. Introduction

* Corresponding author at: Department of Epidemiology, University of Michigan, 1415 Washington Heights SPH-II 5533, Ann Arbor, MI 48109-2029, United States. ** Corresponding author. E-mail addresses: [email protected] (S.D. Mitro), [email protected] (L.S. Rozek), [email protected] (P. Vatanasapt), [email protected] (K. Suwanrungruang), [email protected] (I. Chitapanarux), [email protected] (S. Srisukho), [email protected], [email protected] (H. Sriplung), [email protected] (R. Meza). 1 Department of Environmental and Occupational Health, Milken Institute School of Public Health at the George Washington University, Washington, DC, United States (current affiliation). http://dx.doi.org/10.1016/j.canep.2016.07.002 1877-7821/ã 2016 Elsevier Ltd. All rights reserved.

Thyroid cancer incidence is increasing worldwide [1,2]. In particular, it is among the most common cancers in the U.S., where incidence has increased dramatically over the last 35 years [1–3]. This increase has been largely driven by papillary thyroid cancers, which tripled in incidence in females between 1973 and 2010 and have been increasing at a rate of 6–7% a year since 1992 [3,4]. Papillary tumors are increasing more rapidly in females than in males in the U.S., and small papillary tumors are increasing more quickly than larger ones [1–3]. Because the incidence rate of thyroid cancer is more than twice as high in developed than developing countries, and because

S.D. Mitro et al. / Cancer Epidemiology 43 (2016) 92–99

thyroid cancer mortality has remained stable despite increasing incidence [5–7], some theories attribute the rising incidence to increased screening or variations in diagnosis [1,3,8]. However, these theories are unlikely to fully explain international increases in thyroid cancer. Additionally, increases in large tumors and rising overall thyroid cancer incidence have also been observed in groups traditionally underserved by screening in the U.S. [3,4]. Therefore, unidentified risk factors are hypothesized to be contributing to the increase [2,4]. Exposure to ionizing radiation in childhood is the only widely accepted risk factor for thyroid cancer [8–10]. Some evidence has associated thyroid cancer risk with a variety of other factors, including family history, dietary nitrates, obesity, diabetes, and physical activity, but none has been conclusively shown to elevate risk [11–14]. Iodine deficiency is a potential risk factor for thyroid cancer. Goiter, which is commonly caused by iodine deficiency, is strongly associated with thyroid cancer risk [8]. Iodine deficiency appears to particularly elevate risk of follicular cancers, and residence in an endemic goiter region is associated with increased risk of follicular carcinoma [15,16]. Interestingly, some data suggest that high iodine intake does not reduce overall thyroid cancer risk and may even elevate risk of papillary carcinomas [8,17]. However, the evidence linking iodine supplementation to increases in papillary carcinomas is mixed. Though a number of country-level studies have reported an increase in the papillary:follicular carcinoma incidence ratio as well as an increase in papillary cancer incidence after the implementation of national iodine supplementation, a similar increasing ratio has also been observed in countries that did not introduce iodine supplementation [15]. Thailand presents a natural experiment to study the effects of iodine on thyroid cancer because of its heterogeneous history of iodine deficiency and recent iodine supplementation. Iodine deficiency, approximated by goiter prevalence, has historically been high in Thailand, particularly in the “Goiter Belt” of the north and northeast [18,19]. Studies in the 1950s and 60s reported high levels of endemic goiter and low urinary iodine in the northern, but not southern, regions [19,20]. Because iodine deficiency impairs cognitive development, the Thai government introduced iodized salt programs in the affected areas between 1965 and 1969, but distribution problems and decreased funding after the program’s initial success led to a return of severe deficiency in some areas by the 1980s [19,21,22]. In response, in 1989 the Thai government began small-scale iodine supplementation programs in target provinces. From 1991–1995 iodized salt was again widely distributed, leading to a substantially reduced national prevalence of iodine deficiency and goiter [19]. Between 1993 and 1999, the total goiter prevalence in the north and northeast fell from 10 to 12% to around 3%, while in the south goiter prevalence fell from 4% to <1% of the population [18]. However, after 1999 governmental interest in iodine deficiency lessened. Between 2000 and 2005, median urinary iodine in pregnant women steadily declined below sufficient levels in north Thailand, while remaining below sufficient in northeast Thailand [18,23]. By 2005 median urinary iodine in pregnant women across15 target provinces again declined below sufficient levels [19], and deficiency was much more pronounced in the northern regions: 20–40% of Chiang Mai Province neonates had abnormal TSH in 2005–2006 [24], while only about 9% of neonates in Songkhla Province did in 2006–2007 [25]. Research in the early 1990s indicated that although papillary carcinoma was the most prevalent histological type in most of Thailand, follicular carcinoma was the most common histological type in Khon Kaen province (northeast), possibly due to the higher prevalence of iodine deficiency and goiter [26]. Interestingly, a recent prospective study in Khon Kaen, analyzing 17 cases

93

diagnosed between 1990 and 2011, recorded more papillary than follicular cancers [27]. A separate cross-sectional study comparing regional registries confirmed that thyroid cancer incidence varies regionally, and the Songkhla province registry (in the south) recorded the highest thyroid cancer rates nationally in 2007 [28]. In these analyses, we examined trends in thyroid cancer in Thailand from 1990 to 2009 in three geographically separated Thai populations (Songkhla Province in the south, Chiang Mai Province in the north, and Khon Kaen Province in the northeast, Fig. 1). We hypothesized that papillary cancer incidence would increase, and follicular cancer incidence would decrease, as iodine deficiency decreased nationally in Thailand. Furthermore, we expected these trends to be particularly pronounced in areas known to be historically iodine-deficient. Finally, we examined the timing of histological trends in light of periods of national iodine supplementation, to assess whether supplementation affected population-level rates of thyroid cancer. 2. Materials and methods 2.1. Registries and case ascertainment The Songkhla Registry was established in 1989 and actively extracts case data from community and private hospitals, the provincial health office, and the provincial population registration office. The Chiang Mai Registry was established in 1983 and actively extracts case data from all province public and private hospitals, as well as medical and pathology clinics [29]. The Khon Kaen Registry was established in 1987 and actively extracts case data from private and government hospitals, as well as health centers [30]. All three registries have been included in the International Agency for Research on Cancer’s publication, Cancer

Fig. 1. Songkhla Province (1), Chiang Mai Province (2), and Khon Kaen Province (3), Thailand. Base map of Thailand created by Wikipedia contributor Ahoerstemeier.

94

S.D. Mitro et al. / Cancer Epidemiology 43 (2016) 92–99

Incidence in Five Continents. Songkhla Registry was included in Volumes 8–10; Khon Kaen Registry in Volumes 6–8, and 10; and Chiang Mai Registry in Volumes 6–10. Death certificate only registrations were less than 4% of cases in all registries, across all years. Most cases were microscopically verified (84–96% in Chiang Mai; 70–98% in Khon Kaen; 96–99% in Songkhla, varying by year). Completeness was over 90% by capture-recapture methods [31]. Thyroid cancer cases were extracted from all three registries for years 1990–2009, using ICD-10 code C73. Extracted information included age, sex, date of diagnosis, and tumor histological subtype. Person-years were calculated from province-specific population and housing census data collected by Thailand’s National Statistics Office [32–34]. Cases were divided by histological subtype into four groups: papillary carcinoma (ICD-O-3 codes 8050, 8260, 8340, 8341), follicular carcinoma (8290, 8330, 8331, 8335), medullary or anaplastic carcinoma (8020, 8021, 8030, 8031, 8032, 8345, 8510) and other carcinomas (8000, 8001, 8010, 8012, 8041, 8052, 8070, 8071, 8072, 8073, 8074, 8140, 8310, 8430, 8830, 8890, 9120). 2.2. Trend analysis Annual age-adjusted thyroid cancer incidence rates in each registry were standardized to the Segi (1960) world population to maintain comparability with other international data [35]. Age-adjusted incidence was calculated for males and females overall, and for each histological type and gender combination. Trends in age-adjusted incidence over time were calculated using Joinpoint Regression Program 4.0.4 to find annual percent change in each trend segment and to locate statistically significant trend change points [36]. Analyses were conducted for each registry and then for the three registries combined. 2.3. Age-Period-Cohort analysis Age-period-cohort (APC) analyses were performed to assess the separate effects of age at diagnosis, year of diagnosis, and birth cohort, on thyroid cancer incidence over time. Age-specific incidence rates were calculated for 17 age groups (5–9, 10–14, . . . , 80–84, 85) and for 20 calendar years (1990–2009 in 1-year intervals) for both sexes. Because APC models containing age, period, and cohort variables are inherently non-identifiable, we fitted 2 two-effects models (age-period, AP, and age-cohort, AC) and compared the residuals from each model to determine whether year of birth or year of diagnosis better

correlated with thyroid cancer incidence trends [3,37]. After fitting the two-effect models, we fitted the remaining effect (cohort in AP models, and period in AC models) to the residuals (AP-C or AC-P models). APC analyses were performed in R version 3.0.1 using the Epi package. 3. Results 3.1. Descriptive Between 1990–2009, there were 942 cases recorded in Songkhla Registry (77.7% female), 959 cases in Khon Kaen Registry (82.8% female), and 848 cases in Chiang Mai Registry (78.2% female). In all three registries, females were significantly more likely than males to be diagnosed with thyroid cancer, and were significantly younger at diagnosis than males (p < 0.01) (not shown). Histological types of thyroid cancer were diagnosed at different rates in each registry (Table 1). Papillary carcinomas were diagnosed at a higher rate in both sexes in Songkhla province than in other registries, while medullary or anaplastic cancers were diagnosed at a higher rate in both sexes in Chiang Mai (p < 0.05). 3.2. Trends 3.2.1. Sex After pooling the registries, neither sex exhibited a significant incidence trend over 1990–2009 in overall cancer incidence (all cancer subtypes combined; Male APC: 1.4; female APC: 0.9) (Fig. 2, Table 2), though trends varied somewhat by registry. Cancer incidence significantly increased over time in Songkhla females (APC: 3.6 [1990–2002, p < 0.05], 16.4 [2002–2005], 16.2 [2005–2009, p < 0.05]) but the trend was not significant in Chiang Mai females (APC:-0.1) or Khon Kaen females (APC: 1.7 [1990–2009]) (Fig. 2, Table 2). Cancer incidence significantly decreased over time in Chiang Mai males (APC: 4.6, p < 0.05), but the trend was not significant in Songkhla males (APC: 0.1) or Khon Kaen males (APC: 0.4) (Fig. 2, Table 2). 3.2.2. Histology Papillary carcinoma incidence significantly increased from 1990 to 2001 in all females (APC: 7.3 [1990–2001, p < 0.05], 2.1 [2001–2009]) and in all males (APC: 2.0, [1990–2009, p < 0.05]) after pooling (Fig. 3, Table 2). Papillary carcinomas increased significantly in Chiang Mai females (3.9, p < 0.05) and Khon Kaen females (APC: 6.5, p < 0.05), and increased moderately

Table 1 Clinical characteristics of thyroid cancer cases in females and males across registries, 1990–2009. Rates are age-adjusted to the Segi world population and calculated per 100,000 person-years. Total

Songkhla

Chiang Mai

Khon Kaen

Characteristics

N

Rate

N

Rate

N

Rate

N

Rate (95% CI)

Average Population: Females Average Population: Males Total Cases: Females Histology: Females Papillary Follicular Medullary/Anaplastic Other

2,243,161 2,194,267 2189

4.37 (4.19, 4.56)

637,343 614,763 732

5.39 (4.99, 5.79)

751,489 741,963 663

3.75 (3.45, 4.03)

854,329 837,541 794

4.20 (3.91, 4.50)

1344 513 93 239

2.64 (2.50, 2.78) 1.03 (0.94, 1.12) 0.20 (0.16, 0.24) 0.50 (0.43, 0.56)

493 173 19 47

3.60 (3.28, 3.92) 1.28 (1.09, 1.47) 0.15 (0.08, 0.22) 0.36 (0.26, 0.47)

375 168 51 69

2.08 (1.87, 2.30) 0.95 (0.81, 1.10) 0.30 (0.22, 0.39) 0.41 (0.31, 0.50)

476 172 23 123

2.47 (2.25, 2.70) 0.92 (0.78, 1.06) 0.13 (0.08, 0.18) 0.68 (0.56, 0.80)

Total Cases: Males Histology: Males Papillary Follicular Medullary/Anaplastic Other

560

1.25 (1.14, 1.35)

210

1.74 (1.50, 1.98)

185

1.13 (0.96, 1.29)

165

1.00 (0.85, 1.16)

308 120 55 77

0.67 (0.59, 0.74) 0.27 (0.22, 0.32) 0.13 (0.09, 0.16) 0.18 (0.14, 0.22)

142 38 10 20

1.17 (0.98, 1.37) 0.31 (0.21, 0.41) 0.08 (0.03, 0.13) 0.18 (0.10, 0.26)

76 50 28 31

0.45 (0.35, 0.55) 0.31 (0.22, 0.39) 0.18 (0.11, 0.24) 0.20 (0.13, 0.27)

90 32 17 26

0.52 (0.41, 0.63) 0.20 (0.13, 0.28) 0.11 (0.06, 0.16) 0.17 (0.10, 0.23)

●

●

2

●

● ●

● ● ●

● ●

●

●

1990

● ● ● ● ● ● ● ●

● ● ●

1995

2000

8 6

●

●

●

b. ● ● ●

●

4

●

●

●

●

●

4

●

●

●

●

● ● ●

● ●

2005

● ● ● ● ●

1995

● ●

●

●

●

2

●

1990

●

●

● ●

● ● ● ● ●

●

1995

2000

●

●

●

2005

●

●

2000

●

● ●

●

2005

8

d.

6 4

● ●

● ● ● ●

●

● ●

● ●

●

● ● ●

●

● ●

● ●

● ●

● ● ●

●

●

2

●

● ●

●

●

●

● ● ● ● ● ● ●

●

● ● ● ●

●

● ● ●

● ● ●

●

0

●

Age−adjusted Rate per 100,000

8 6 4

● ●

●

0

Age−adjusted Rate per 100,000

● ●

● ●

●

Year

● ●

● ●

●

1990

c.

●

●

●

● ●

●

●

●

● ●

Year

●

●

●

2

6

● ●

●

●

95

0

●

●

Age−adjusted Rate per 100,000

8

a.

0

Age−adjusted Rate per 100,000

S.D. Mitro et al. / Cancer Epidemiology 43 (2016) 92–99

1990

1995

Year

2000

2005

Year

Fig. 2. Sex-specific rates of thyroid cancer in a. Songkhla, b. Chiang Mai, and c. Khon Kaen registries, 1990–2009, for all cancer subtypes combined. d. Sex-specific rates of thyroid cancer pooled across registries. Rates were age-adjusted to the Segi world population. Black dots represent observed yearly rates in females; red lines represent modeled yearly rates in females. Gray dots represent observed yearly rates in males; blue lines represent modeled yearly rates in males. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 2 Annual percent change (APC) and 95% confidence interval (CI) in age-adjusted incidence, 1990-2009. Trend 1

Trend 2

Trend 3

APC (95% CI)

Years

APC (95% CI)

Females All females Papillary Follicular Medullary/Anaplastic Other Songkhla Papillary Follicular Khon Kaen Papillary Follicular Other Chiang Mai Papillary Follicular

0.9 ( 0.3, 2.1) 7.3 (4.6, 10.1)* 4.3 ( 7.8, 0.8)* 0.5 ( 3.6, 4.7) 2.0 ( 5.7, 1.9) 3.6 (1.4, 5.9)* 2.0 ( 0.4, 4.5) 0.8 ( 4.2, 2.8) 1.7 ( 0.5, 4.0) 6.5 (2.3, 10.9)* 4.5 ( 7.0, 2.0)* 1.8 ( 5.7, 2.3) 0.1 ( 1.7, 1.6) 3.9 (1.2, 6.7)* 3.5 ( 5.8, 1.2)*

1990–2009 1990–2001 1990–98 1990–2009 1990–2009 1990–2002 1990–2009 1990–2009 1990–2009 1990–2009 1990–2009 1990–2009 1990–2009 1990–2009 1990–2009

– 2.1 ( 6.0, 2.1) 12.3 ( 19.8, 57.2) – – 16.4 ( 42.0, 20.3) – – – – – – – – –

– 2001–09 1998–2001 – – 2002–05 – – – – – – – – –

– –

Males All males Papillary Follicular Other Songkhla Papillary Khon Kaen Chiang Mai Papillary

1.4 ( 2.9, 0.1) 2.0 (0.2, 3.9)* 5.2 ( 9.3, 1.0)* 4.7 ( 8.3, 1.0)* 0.1 ( 3.0, 2.9) 0.8 ( 2.6, 4.2) 0.4 ( 4.5, 3.8) 4.6 ( 8.2, 0.9)* 0.2 ( 5.4, 5.1)

1990–2009 1990–2009 1990–2009 1990–2009 1990–2009 1990–2009 1990–2009 1990–2009 1990–2009

– – – – – – – – –

– – – – – – – – –

– – – – – – – – –

*

P < 0.05.

Years

Trend 4

APC (95% CI)

17.0 ( 29.8, 1.8) – – 16.2 (3.5, 30.4)* – – – – – – – – –

*

Years

APC (95% CI)

Years

– – 2001–05 – – 2005–09 – – – – – – – – –

– – 8.2 ( 2.7, 20.4) – – – – – – – – – – – –

– – 2005–09 – – – – – – – – – – – –

– – – – – – – – –

– – – – – – – – –

– – – – – – – – –

96

●

2

●

●

●

●

●

● ●

●

● ● ● ●

●

● ●

●

●

●

●

●

● ●

5

● ● ● ●

● ●

● ●

●

● ●

●

●

● ●

●

● ● ●

● ● ●

● ●

●

● ● ● ● ● ●

●

●

●

●

● ● ●

●

4

● ● ● ●

3

4

●

● ●

●

●

●

● ●

●

2

●

●

c

●

●

● ●

● ●

1

3

●

●

Age−adjusted Rate per 100,000

●

3

● ● ●

2

4

● ●

●

b

1

●

●

1

Age−adjusted Rate per 100,000

● ●

Age−adjusted Rate per 100,000

a

●

●

5

5

S.D. Mitro et al. / Cancer Epidemiology 43 (2016) 92–99

●

● ● ●

● ●

●

● ● ●

●

2000

2005

1990

1995

3 2

●

●

●

● ●

● ● ●

●

●

1

● ●

●

● ●

●

●

●

● ●

● ● ●

● ●

●

●

1995

2000

2000

2005

Year

●

e

2005

●

● ●

● ●

●

●

●

●

● ●

●

●

●

●

●

● ●

●

●

●

●

● ●

● ●

●

● ●

●

Year

● ●

●

● ●

1990

●

● ●

0.0

1990

1995

1.5 ●

●

●

● ●

1990

1.0

● ●

●

2005

0.5

● ● ●

Age−adjusted Rate per 100,000

4

d

0

Age−adjusted Rate per 100,000

2000 Year

5

Year

●

● ●

●

0

0

0

1995

●

●

●

1990

●

● ●

1995

2000

● ●

2005

Year

Fig. 3. Histology-specific rates of thyroid cancer per 100,000 population in a. Songkhla; b. Chiang Mai; and c. Khon Kaen registries, 1990–2009. d. Histology-specific rates of thyroid cancer per 100,000 pooled across registries in females. e. Histology-specific rates of thyroid cancer per 100,000 pooled across registries in males. Rates were ageadjusted to the Segi world population. Panel e. uses a different scale due to the much lower incidence rate in men. Black dots represent observed yearly rates of papillary cancer; orange lines represent modeled yearly rates of papillary cancer. Gray dots represent observed yearly rates of follicular cancer; green lines represent modeled yearly rates of follicular cancer. Dotted line at 1991 represents the beginning of national iodination. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

in Songkhla females (APC: 2.0) (Fig. 3, Table 2). No significant trends in papillary carcinomas over time were found for males in any individual registry (Table 2). Conversely, follicular carcinoma incidence significantly decreased over time in both males (APC: 5.2, p < 0.05) and females (-4.3 [1990–1998, p < 0.05], 12.3 [1998–2001], 17.0 [2001–2005, p < 0.05], 8.2 [2005–2009]) after pooling (Fig. 3, Table 2). Among individual registries, follicular carcinomas significantly decreased over time in Khon Kaen females (APC: 4.5, p < 0.05) and Chiang Mai females (APC: 3.5, p < 0.05), but the trend was not significant in Songkhla females (APC: 0.8) (Fig. 3, Table 2). The data were insufficient to test for trends in male follicular cancers by registry. No significant trends were found for medullary or anaplastic cancers in females after pooling (APC: 0.5) (Table 2) and the data were insufficient to test for trends in male medullary or anaplastic cancers even after pooling. Cancers categorized as “Other” significantly decreased in males after pooling (APC: 4.7, p < 0.05), but the negative trend was not significant in females (APC: 2.00) (Table 2).

3.3. Age-period-cohort analysis AP-C and AC-P models were fit for females in each registry, females pooled across registries, and pooled papillary and follicular histological types in females. The number of male cases was insufficient to support adequately powered models. In females, AP models fit the data better than AC models in all registries except Chiang Mai, and also fit the papillary cancer and follicular cancer data better than AC models (Fig. 4). Fig. 4 shows the estimated age, period and cohort effects for all AP-C and AC-P models. The AP-C models show the significant increase by period in the incidence of papillary tumors and the corresponding decrease in the incidence of follicular tumors, consistent with the Joinpoint trend analyses of age-adjusted rates. 4. Discussion Over 1990–2009, the incidence of papillary cancers increased, and the incidence of follicular cancers decreased, in both sexes in

1990

1995

2000

2005

Calendar Year

2000

Papillary AC−P Period Effects

1990

1995

2000

2005

Calendar Year

Follicular AP−C Cohort Effects

1920

1960

2000

Birth Year

Follicular AP−C Period Effects

1990

1995

2000

2005

Calendar Year

0.5

Age−specific Incidence

80

20

40

60

80

Age

Follicular AC−P Cohort Effects 0.5

0.0 2.5

Age−specific Incidence

60

Age

Relative Incidence

1960 Birth Year

40

Follicular AC−P Age Effects

1920

1960

2000

Birth Year

Follicular AC−P Period Effects 0.9

1920

20

0.75

Papillary AC−P Cohort Effects

Relative Incidence

8

80

Follicular AP−C Age Effects

Relative Incidence

Papillary AP−C Period Effects

0

Age−specific Incidence

2000

60

Age

97

0.6

1960 Birth Year

40

Relative Incidence

1920

20

0.5

Papillary AP−C Cohort Effects

Relative Incidence

80

1.2

60

Age

Papillary AC−P Age Effects

0.8

40

Relative Incidence

0.5

20

0.90

Relative Incidence Relative Incidence

Papillary AP−C Age Effects

1.0

Age−specific Incidence

S.D. Mitro et al. / Cancer Epidemiology 43 (2016) 92–99

1990

1995

2000

2005

Calendar Year

Fig. 4. Age-Period models (residuals fitted to Cohort values) in the left column, and Age-Cohort models (residuals fitted to Period values) in the right column for pooled females. Follicular and papillary carcinoma trends are graphed. Age-effects represent the age-specific incidence (per 100,000); Cohort effects represent relative incidence with respect to the 1950 birth-cohort; period effects represent relative incidence with respect to 1992.

6 4 2 0

Age−adjusted rate ratio per 100,000

8

Thailand (based on three pooled cancer registries). As anticipated, the largest increase in papillary cancers, and the largest decrease in follicular cancers, was seen in regions known to be historically iodine-deficient (Chiang Mai and Khon Kaen). Songkhla, which historically has had lower levels of iodine deficiency, did not exhibit a significant increase in papillary carcinomas or a significant decrease in follicular carcinomas over 1990–2009 (Fig. 3). The heterogeneity in thyroid cancer incidence trends by region and histology in Thailand are suggestive of an iodination effect. Severe iodine deficiency persisted in Thailand until the national initiative to reduce iodine deficiency in the 1990s. To examine changes in carcinogenesis over time, we compared the ratio of papillary:follicular cancers from 1990 to 2009. The ratio of papillary:follicular cancers was close to 1:1 for both males and females at the beginning of registry data collection in 1990, but

1990

1995

2000

2005

Year

Fig. 5. Papillary: Follicular rate ratio, 1990–2009. The black line represents the rate ratio in females. The gray line represents the rate ratio in males. Rates were ageadjusted to the Segi world population.

increased to 3–4:1 by 2009 (Fig. 5), suggesting an effect on carcinogenesis of the iodination initiative. Interestingly, the significant increase in papillary cancer in females (1990–2001) and the significant national decrease in follicular cancer in females (1990–1998) across the three pooled registries also coincided with Thailand’s most recent national iodination push (1989–1999) [19]. Due to the ecologic study design, we cannot determine whether the observed incidence changes were caused by iodination or simply occurred at the same time, though the correlation is suggestive of an association. Iodine’s potential role in thyroid carcinogenesis has not been fully investigated, and the time needed to observe an effect of iodine supplementation on thyroid cancer incidence remains unknown. Some studies have suggested that effects may be evident soon after the onset of iodination, affecting later stages of carcinogenesis, while others suggest that effects would be observed 15–20 years after implementation, affecting earlier stages [38,39]. Nationwide salt iodination has coincided with an increase in thyroid cancer in several countries, but it is difficult to untangle the relative effects of iodination and increased surveillance. Indeed, a study in Denmark quantifying the effect of a 2000 national iodination policy noted that papillary cancer incidence had actually begun increasing a few years before the iodination policy [40]. Italy, which has no iodination policy, also recorded an increase in papillary cancers between 1991 and 2005 [41]. No significant change in overall thyroid cancer incidence was seen in either sex when the registries were pooled. However, a significant increase in thyroid cancer incidence was observed in Songkhla females, and a significant decrease was observed in Chiang Mai males. Previous studies associating iodination with thyroid cancer have been mixed, with some suggesting that iodination increases overall cancer, and others finding no significant overall increase [15,38,39]. APC analysis indicates that the observed incidence trends in females are better explained by Period than Cohort effects. Other APC studies of thyroid cancer have produced mixed results. Some have found that incidence over time may be better explained by Cohort, not Period effects, perhaps due to historical medical practices such as radiation therapy for children producing large cohort-specific effects [40,42]. Others have found that AP effects explained the incidence better than AC effects, possibly because of the recent effects of increased screening and more precise cancer

98

S.D. Mitro et al. / Cancer Epidemiology 43 (2016) 92–99

detection techniques [3,43]. Increased screening is unlikely to explain thyroid cancer trends in Thailand over the study period, especially because no overall increase in thyroid cancer diagnosis was observed across the three registries after pooling. Instead, a different period-specific factor, such as iodine supplementation, may explain the effect observed in females. This study has several strengths. It is the first to directly compare three geographically separate Thai cancer registries to report long-term changes in thyroid cancer incidence and histology over time. It adds to a currently sparse literature documenting thyroid cancer trends in Southeast Asia. Additionally, this analysis used population-based registries each containing several decades of data, totaling several thousand cases of thyroid cancer. Finally, Thailand provides a natural experiment, because iodine deficiency has varied greatly by region and time. We were thus able to measure thyroid cancer trends during and after a national push to increase iodine sufficiency in historically deficient regions, allowing for an examination of immediate effects of iodine supplementation. This study also has weaknesses. Thailand’s cancer registries were established relatively recently, preventing examination of region-specific cancer rates and histological patterns prior to national iodine supplementation. Nonetheless, the registries span 20 years, allowing for a detailed analysis of recent trends. Registry data did not include information on other relevant risk factors that may vary regionally, including socioeconomic status, obesity, diet, radiation exposure, and family history, limiting the assessment of alternative risk factors. Although we could not rule out increasing access to care as a contributing factor to increased diagnosis, the medical procedures and treatment facilities for thyroid cancer are very similar among the 3 provinces. Additionally, the quality of surveillance was similar in the 3 provinces as the registries worked as a network to ensure standard practices. Few male cases were recorded in each registry, limiting registryspecific male histology analyses. However, pooling the registries provided sufficient cases for a more detailed analysis of male thyroid cancer trends. The registries did not collect stage or tumor size for the majority of cases, preventing examination of clinical differences between regions. Pooling the registries does not create a nationally representative sample of the Thai population. We were unable to explain the lower rate of cancer seen in between 2000 and 2005. Around the year 2000, a change in the national health insurance scheme created an electronic population database for every province, which may have affected completeness or accuracy for residency records in some provinces. Finally, some cases recorded in each registry may have occurred in people who traveled to one of the provinces to be treated, which may have altered the recorded incidence rate. However, these cases are probably less than 5% of all cases and are not expected to have affected the overall results. In conclusion, this study adds to the research indicating that papillary carcinoma incidence increases, and follicular carcinoma incidence decreases, as population-level iodine deficiency declines. Additionally, the yearly trends in incidence reported here suggest that the effect of iodine supplementation on carcinoma histology may be immediate. However, these results should be interpreted with caution, as our findings are limited by the ecological study design and lack of data prior to iodine supplementation. Future research should further examine the direct effects of iodine supplementation on thyroid cancer incidence in other regions and settings. Conflicts of interest None.

Authors’ contribution PV, KS, IC, SS, and HS collected the data and prepared the data for analysis. RM, LR, HS, and SM designed the analysis. SM conducted the analysis. RM, LR, HS, and SM interpreted the findings. SM drafted the manuscript. All authors revised the manuscript for intellectual content. Acknowledgements Funding provided by: University of Michigan School of Public Health Office of Global Public Health and the University of Michigan Center for Southeast Asian Studies, and the Thailand National Science and Technology Development Agency (P-10-10307). The funding sources had no role in the in study design; collection, analysis, and interpretation of data; writing of the report; and the decision to submit the paper for publication. References [1] L. Davies, H.G. Welch, Increasing incidence of thyroid cancer in the United States, 1973–2002, JAMA J. Am. Med. Assoc. 295 (2006) 2164–2167. [2] B.A. Kilfoy, T. Zheng, T. Holford, et al., International patterns and trends in thyroid cancer incidence 1973–2002, Cancer Cause Control 20 (2009) 525–531. [3] R. Meza, J.T. Chang, Multistage carcinogenesis and the incidence of thyroid cancer in the US by sex, race, stage and histology, BMC Publ. Health 15 (2015) 789. [4] B. Aschebrook-Kilfoy, E.L. Kaplan, B.C. Chiu, P. Angelos, R.H. Grogan, The acceleration in papillary thyroid cancer incidence rates is similar among racial and ethnic groups in the United States, Ann. Surg. Oncol. 20 (2013) 2746–2753. [5] H.S. Ahn, H.J. Kim, H.G. Welch, Korea’s thyroid-cancer epidemic—screening and overdiagnosis, N. Engl. J. Med. 371 (19) (2014) 1765–1767. [6] L. Davies, L.G. Morris, M. Haymart, et al., American Association of Clinical Endocrinologists and American College of Endocrinology disease state clinical review: the increasing incidence of thyroid cancer, Endocr. Pract. 21 (6) (2015) 686–696. [7] S. Vaccarella, L. Dal Maso, M. Laversanne, et al., The impact of diagnostic changes on the rise in thyroid cancer incidence: a population based study in selected high-resource countries, Thyroid 25 (10) (2015) 1127–1136. [8] L. Dal Maso, C. Bosetti, C. La Vecchia, S. Franceschi, Risk factors for thyroid cancer: an epidemiological review focused on nutritional factors, Cancer Cause Control 20 (2009) 75–86. [9] E. Ron, J.H. Lubin, R.E. Shore, et al., Thyroid cancer after exposure to external radiation: a pooled analysis of seven studies, Radiat. Res. 141 (1995) 259–277. [10] E. Cardis, A. Kesminiene, V. Ivanov, et al., Risk of thyroid cancer after exposure to 131I in childhood, J. Natl. Cancer I 97 (2005) 724–732. [11] G.M. Oakley, K. Curtin, R. Pimental, L. Buchmann, J. Hunt, Establishing a familial basis for papillary thyroid carcinoma using the Utah Population Database, JAMA Otolaryngol. Head Neck Surg. 139 (2013) 1171–1174. [12] M.H. Ward, B.A. Kilfoy, P.J. Weyer, K.E. Anderson, A.R. Folson, J.R. Cerhan, Nitrate intake and the risk of thyroid cancer and thyroid disease, Epidemiology 21 (2010) 389–395. [13] C.M. Kitahara, E.A. Platz, L.E. Beane Freeman, et al., Obesity and thyroid cancer risk among U.S. men and women: a pooled analysis of five prospective studies, Cancer Epidemiol. Biomark. 20 (2011) 464–472. [14] D. Schmid, G. Behrens, C. Jochem, M. Keimling, M. Leitzmann, Physical activity, diabetes, and risk of thyroid cancer: a systematic review and meta-analysis, Eur. J. Epidemiol. 28 (2013) 945–958. [15] H.R. Harach, G.A. Ceballos, Thyroid cancer, thyroiditis and dietary iodine: a review based on the Salta, Argentina model, Endocr. Pathol. 19 (2008) 209– 220. [16] M.R. Galanti, P. Sparen, A. Karlsson, L. Grimelius, A. Ekbom, Is residence in areas of endemic goiter a risk factor for thyroid cancer? Int. J. Cancer 61 (1995) 615– 621. [17] T. Sehestedt, N. Knudsen, H. Perrild, C. Johansen, Iodine intake and incidence of thyroid cancer in Denmark, Clin. Endocrinol. 65 (2006) 229–233. [18] The World Health Organization Vitamin and Mineral Nutrition Information System (VMNIS), Database on Iodine Deficiency: Thailand. Available at http:// who.int/vmnis/iodine/data/database/countries/tha_idd.pdf?ua=1 (accessed 30.01.14). [19] S. Sinawat, Iodine deficiency disorders in Thailand, in: V.R. Preedy, G.N. Burrow, R.R. Watson (Eds.), Comprehensive Handbook of Iodine: Nutritional, Biochemical, Pathological, and Therapeutic Aspects, Elsevier Science BV, Amsterdam, 2009, pp. 1221–1226. [20] R.H. Follis, K. Vanprapa, D. Damrongsakdi, Studies on iodine nutrition in Thailand, J. Nutr. 76 (1962) 159–173. [21] The World Health Organization, Nutrition: Micronutrient deficiencies. Available at http://www.who.int/nutrition/topics/idd/en/ (accessed 16.02.15).

S.D. Mitro et al. / Cancer Epidemiology 43 (2016) 92–99 [22] R. Pleehachinda, Urinary iodine excretion and thyroid function studies in endemic goiter in northern Thailand, J. Med. Assoc. Thail. 67 (1984) 31–35. [23] R. Rajatanavin, Iodine deficiency in pregnant women and neonates in Thailand, Publ. Health Nutr. 10 (12A) (2007) 1602–1605. [24] W. Charoensiriwatana, P. Srijantr, N. Janejai, S. Hasan, Application of geographic information system in TSH neonatal screening for monitoring of iodine deficiency areas in Thailand, Southeast Asian J. Trop. Med. Publ. Health 39 (2) (2008) 362–367. [25] S. Jaruratanasirikul, P. Sangsupawanich, O. Koranantakul, et al., Maternal iodine status and neonatal thyroid-stimulating hormone concentration: a community survey in Songkhla, southern Thailand, Publ. Health Nutr. 12 (12) (2009) 2279–2284. [26] V. Vatanasapt, S. Sriamporn, P. Vatanasapt, Cancer control in Thailand, Jpn. J. Clin. Oncol. 32 (Supplement 1) (2002) S82–91. [27] W. Sungwalee, P. Vatanasapt, S. Kamsa-ard, K. Suwanrunruang, S. Promthet, Reproductive risk factors for thyroid cancer: a prospective cohort study in Khon Kaen, Thailand, Asian Pac. J. Cancer Prev. 14 (2013) 5153–5155. [28] M.A. Moore, P. Attasara, T. Khuhaprema, et al., Cancer epidemiology in mainland South-East Asia—past, present, and future, Asian Pac. J. Cancer P 9 (2008) 67–80. [29] Faculty of Medicine, Chiang Mai University. Cancer Incidence and Mortality in Chiang Mai, 2007. Chiang Mai, Thailand: Chiang Mai Cancer Registry, 2010. [30] V. Vatanasapt, S. Sriamporn, S. Kamsaard, K. Suwanrungruang, P. Pengsaa, D.J. Charoensiri, J. Chaiyakum, M. Pesee, Cancer survival in Khon Kaen, Thailand, IARC Sci. Publ. 145 (1998) 123–134. [31] K. Suwanrungruang, H. Sriplung, P. Attasara, et al., Quality of case ascertainment in cancer registries: a proposal for a virtual three-source capture-recapture technique, Asian Pac. J. Cancer Prev. 12 (1) (2011) 173–178. [32] National Statistical Office, 1990 Population and Housing Census. Office of the Prime Minister: Bangkok. 1994.

99

[33] National Statistical Office, 2000 Population and Housing Census. Office of the Prime Minister: Bangkok. 2002. [34] National Statistic Office, 2010 Population and Housing Census. Office of the Prime Minister: Bangkok. 2013. [35] F. Bray, A. Guilloux, R. Sankila, D.M. Parkin, Practical implications of imposing a new world standard population, Cancer Causes Control 13 (2002) 175–182. [36] Joinpoint Regression Program, Version 4.0.4—May 2013: Statistical Methodology and Applications Branch, Surveillance Research Program, National Cancer Institute. [37] T.R. Holford, Understanding the effects of age, period, and cohort on incidence and mortality rates, Annu. Rev. Publ. Health 12 (1991) 425–457. [38] W. Dong, H. Zhang, P. Zhang, et al., The changing incidence of thyroid carcinoma in Shenyang, China before and after universal iodization, Med. Sci. Monit. 19 (2013) 49–53. [39] J. Farahati, M. Geling, U. Mader, et al., Changing trends of incidence and prognosis of thyroid carcinoma in Lower Franconia, Germany, from 1981– 1995, Thyroid 14 (2004) 141–147. [40] M. Blomberg, U. Feldt-Rasmussen, K.K. Andersen, S.K. Kjaer, Thyroid cancer in Denmark 1943–2008, before and after iodine supplementation, Int. J. Cancer 131 (2012) 2360–2366. [41] M. Lise, S. Franceschi, C. Buzzoni, et al., Changes in the incidence of thyroid cancer between 1991 and 2005 in Italy: a geographical analysis, Thyroid 22 (2012) 27–34. [42] T. Zheng, T.R. Holford, Y. Chen, et al., Time trend and age-period-cohort effect on incidence of thyroid cancer in Connecticut, 1935–1992, Int. J. Cancer 67 (1996) 504–509. [43] L. Dal Maso, M. Lise, P. Zombon, et al., Incidence of thyroid cancer in Italy, 1991– 2005: time trends and age-period-cohort effects, Ann. Oncol. 22 (2011) 957– 963.