Seaweed intake and Thyroid cancer risk

class: center, middle, inverse, title-slide

# Seaweed intake and Thyroid cancer risk
### Chaochen Wang| 王　超辰 Aichi Medical University
### 2021-01-14/15 @HOME via zoom

---

class: middle, inverse, center

# Background

---
class: middle

## Background (1)

.small[.bold[
- Iodine is an essential micronutrient that is oxidized to produce iodine-containing thyroid hormones.

- Iodine deficiency has long been a recognized global public health problem, however, there was not much emphasis on the possible "consequences" of **iodine excess**.

- The sources of excess iodine can be from 
  - iodized salt, ~~drinking water~~, ~~milk rich in iodine~~, seaweeds, and dietary supplements.
  
- Iodine excess can cause subclinical or overt thyroid dysfunction, and even thyroid (papillary) cancer.
]]

???

- The recommended dietary allowance (RDA) for adults is 150 micrograms/day, 220 to 250 micrograms/day for pregnant women, and 250 to 290 micrograms/day for breastfeeding women.

- up to 1 mg/day is (probably) safe for most people

---
class: middle

## Background (2)

.small[.bold[
- Thyroid cancer might have already become **the third leading cancer diagnosis** in the United States in women.

- Age-adjusted incidence rate in Japan (2016,  per 100,000 person-years): 
    - 5.7 in men
    - 16.7 in women

- However, the numbers may include substantial portions of "incidentally-detected" cases through health/cancer screenings.

- The only mostly accepted risk factor for thyroid cancer is **radiation exposure**.
]
]
???

- Because the advance of computer tomography, ultrasonography and fine needle aspiration cytology has made it possible to detect small thyroid cancers.

---
class: top, center, inverse
background-image: url("./fig/incidenceratesJapan.png")
background-position: 50% 50%
background-size: contain

???

- The incidence rate of thyroid cancer increased from about 2 to nearly 10 per 100,000 people in Japan over the past three decades. 
- However, mortality rates in both men and women stayed stable or slightly decreased.

---
class: top, center, inverse
background-image: url("./fig/globalincidence.png")
background-position: 50% 50%
background-size: contain

???
https://gco.iarc.fr/today/online-analysis-map?v=2020&mode=population&mode_population=continents&population=900&populations=900&key=asr&sex=0&cancer=32&type=0&statistic=5&prevalence=0&population_group=0&ages_group%5B%5D=0&ages_group%5B%5D=17&nb_items=10&group_cancer=1&include_nmsc=1&include_nmsc_other=1&projection=natural-earth&color_palette=default&map_scale=quantile&map_nb_colors=5&continent=0&rotate=%255B10%252C0%255D

- Global incidence seems to be higher in high-income countries compared with low-income and middle-income countries for both men and women.

---
class: top, center,
background-image: url("./fig/globalmortality.png")
background-position: 50% 50%
background-size: contain

???
- However, mortality in high-income countries are among the lowest worldwide

---
class: top, center,inverse
background-image: url("./fig/metaiodine.png")
background-position: 50% 50%
background-size: contain

???

- One meta analysis which included 16 observational studies reported higher thyroid cancer risk for people with high iodine exposure.

---
class: top
background-image: url("./fig/metaOR.png")
background-position: 50% 10%
background-size: contain

.footnote[.bold[
- This meta-analysis evaluated iodine level measured by several methods and we can see there is **no evidence for increased thyroid cancer risk** in any kind of measurements. 
- Another problem in this study is that they **only included subjects in the highest and the lowest exposure groups**. 
]]

---
class: top
background-image: url("./fig/arsenic.png")
background-position: 50% 10%
background-size: contain

???
- Another review paper published recently also discussed potential adverse effects of seaweed intake, which might caused by both excess iodine and arsenic species.
- Although arsenic has been classified as carcinogenic by the International Agency for Research against Cancer (IARC), but harmful effects on individuals' health would be unlikely unless consume extremely high amounts of seaweed.

---
class: top, center, inverse
background-image: url("./fig/michikawa.png")
background-position: 50% 50%
background-size: contain

???

- There are 2 reports from Japan on this topic, one is from JPHC published in 2011. Where they found daily seaweed consumers had about 71% higher hazard of developing thyroid cancer compared to people had seaweed less than 2 days/week. 
- Furthermore, their stratified analysis reported nearly four fold higher higher hazard in postmenopausal women.

---
class: top, center, inverse
background-image: url("./fig/wang.png")
background-position: 50% 50%
background-size: contain

???

- However, in the JACC study later we published in 2015, we couldn't find the similar association as reported in the JPHC study. 
- We considered that even with increasingly reported incidence over the years, there is still very limited number of cases that could be identified in one or two studies. Another possibility is that we all use different measurements in each study, reference group varies and may have different meanings. We are not sure whether it is appropriate to interpret all the findings in the same way. 
- We need to find a way to conduct proper analysis on each study individually, and find one single maximum effect size for this relationship that we are interested in from each one of the study 
- and then we can use our usual methods to combine these maximum effect size from each study.

---
class: middle

# Objective

.full-width[.content-box-red[.bold[.med[
With greater numbers of participants as well as cases that can be identified through ACC database, we propose to assess **the maximum association** between seaweed intake and the risk of thyroid cancer in ACC for both men and women, through **a revised statistical approach**.
]]]]

---
class: middle, center, inverse

# Methods

---
class: middle

## Methods

.med[.bold[
- Descriptive analysis - show numbers and distributions of: 
    - cases and total participants
    - seaweed intake 
    - Other variables that could related to both exposure and the outcome
- Bayesian surival analysis will be conducted in each joining cohort to estimate an maximum effect size of the relationship.
- Random effect meta-analysis - to evaluate the pooled maximum effect.
]]

---
class: middle

## Methods - Bayesian survival analysis

.med[.bold[
- We already know that incidence increases over time
  - **the hazard ratio could change over time (not proportonal)**
- It is no longer appropriate to use the Cox **proportional** hazard model anymore
- Weibull model in the accelerated failure time (AFT) model form is one option.
  - produce an **acceleration factor** for comparing seaweed consumers to non-consumers about who is **faster/slower** to develop thyroid cancer during a given time period.
]]

---
class: middle

## Methods - Maximum seaweed effect

$$
`\begin{aligned}
\mu_i & = \color{red}{\beta_E}\sum_{j = 0}^{E_i -1} \delta_j + \text{other covariates}\\
\delta & \sim \text{Dirichlet}(\alpha)
\end{aligned}`
$$

.med[.bold[
- Instead of coding the seaweed consumption groups as dummy variables, they should be entered as **ordered categorical predictors** 
 `$\color{red}{\beta_E}$` is the maximum seaweed effect 
 `$\delta_j$` is the fractions (proportions) for each level of seaweed consumption
]]

---
class: middle

## Methods - example

.med[.bold[
- If in A study, the data about seaweed consumption is collected as 5 different levels: ‘never’, ‘1–2 times/month’, ‘1–2 times/week’, ‘3–4 times/week’, and ‘daily or almost daily’ (JACC study). 
- The effect changes from level 0 (never) to level 1 (1-2 times/month) is a proportion `$\delta_0$` of the maximum effect `$\color{red}{\beta_E}$` 
level 1 (1-2 times/month) to level 2 (1–2 times/week) is a proportion `$\delta_1$` of the maximum effect `$\color{red}{\beta_E}$` 
`$\dots$` 
level 3 to level 4 is a proportion `$\delta_4$` of `$\color{red}{\beta_E}$`
- Four fractions `$\delta_0, \delta_1, \dots, \delta_4$`, they sum up to 1: `$\sum_{j=0}^4 = 1$`
]
]

???

- So in this method, we can allow for different exposure measurements in different studies and keep as much information as possible in the model in their original format.
- We could also measure which level contributes the most in the increase of the seaweed effect, if there is any.
- And each study will only produce one single maximum seaweed effect for that study.

---
class: middle

## Methods - combining the `$\color{red}{\beta_E}$`s

.small[.bold[
- These `$\color{red}{\beta_E}$`s are the maximum effects of seaweed on the risk of thyroid cancer.

- Therefore the pooled effect could be estimated, if there are `$K$` studies in the ACC join our working group: 
  - The observed `$\widehat{\beta_{Ek}}, i = 1, \dots, K$` is an estimate of the "real" effect `$\beta_{Ek}$` of study `$k$`
  - These "real" effects `$\beta_{Ek}$` themselves follow a distribution centered with the pooled effect `$\mu$`, and variance `$\tau^2$` - the between-study heterogeneity.
]]

$$
`\begin{aligned}
\widehat{\beta_{Ek}} & \sim N(\beta_{Ek}, \sigma^2_{k}) \\ 
\beta_{Ek} & \sim N(\color{red}{\mu}, \tau^2)
\end{aligned}`
$$

---
class: middle, center, inverse

# Thanks

.bold[.large[link to visit this webslide: 
**[https://wangcc.me/ACC_seaweed/](https://wangcc.me/ACC_seaweed/#1)** 
]]

.bold[.large[Looking forward to seeing your data.]]