# Personal Report in Categorical Date Analysis

## Contents of PPT

### Face

接觸**SV40(猿猴病毒40)** 污染腺病毒疫苗的美國退伍軍人癌症病例對照研究

### Contents
* 摘要
* 動機
* 資料收集方法
* 統計方法
* 結果
* 討論
* 個人想法

### Abstract about this Paper
* 首先SV40是屬於一種疫苗上的意外污染，它於1950∼1960年代在猴子腎臟組織培養中產生。
* 關於SV40，由於在一些實驗室在腫瘤組織中檢測到了SV40的DNA，這顯示SV40可能會導致人類癌症。以此為竊機，展開研究。
* 為了確定被SV40污染的腺病毒疫苗是否與癌症風險增加有關，作者對1959年至1961年入伍的男性退伍軍人進行了 case-control study。

## Introduction of Polio Vaccine and inactivated polio vaccine

我認為這篇paper剛開始的部分有些名詞需要去解釋才能對後來的解讀、理解有更順的幫助。

首先，是關於小兒麻痺疫苗，世界上主要有兩種常見類型
1. 第一種是喬納斯·沙克（Jonas Schack）開發的注射型「沙克疫苗」（IPV），其中含有一劑滅活的小兒麻痺病毒（今天的主角）。
2. 第二種是艾伯特開發的口服「薩賓疫苗」（OPV），薩賓使用減毒小兒麻痺病毒。

接著是如何製作IPV的部分：
1. 使用3种血清型的致病病毒株。
2. 這些病毒首先在一種猴腎組織中培養。
3. 然後用福馬林(主要是甲醛)處理使其失去活性。

## Introduction of Simian Virus 40

接著是關於SV40的詳細介紹
* 如摘要所提到的，SV40是疫苗中意外出現的獼猴病毒於20世紀 50年代末到60世紀初。
* 由於SV40的特性(也就是SV40 存在於滅活小兒麻痺疫苗中)，1955年至1961年間，美國有 1,000萬至3,000萬人（主要是兒童）接種了該疫苗。
* BTW,SV40也存在於其他疫苗中。特別是數十萬名美國新兵提供的腺病毒疫苗。（為什麼？？請參閱下一頁〜）
*

## Motivations of this Study

我們知道何謂SV40後，我們來探討這個Motivations of this Study：
* 我們已知因為多項實驗室證據顯示 V40可能在某些人類惡性腫瘤的病因中發揮作用。在此我們從兩個方向做討論：
1. 理論上：SV40編碼的T抗原(這是一種非結構蛋白)，簡單來說它會結合併**抑制腫瘤抑制蛋白**。
2. 在實驗中（囓齒類）：SV40 疫苗接種會導致各種腫瘤的發生。例如間皮瘤、室管膜瘤、骨肉瘤、白血病…

## Why is the U.S. Military

有了動機後，再來去探討選擇的族群：所以為什麼是美國陸軍?
* 從 1960 年開始，美國陸軍開始為新兵接種腺病毒疫苗，以預防基礎訓練期間==急性呼吸道疾病==的流行。
* 為了更了解與SV40感染相關的潛在癌症風險，我們對 1959 年 1 月至 1961 年 12 月在美國陸軍服役的男性退伍軍人進行了一項癌症病例對照研究。
* 然而，只有少數滅活小兒麻痺疫苗含有活性 SV40，因此有些人接種了腺病毒疫苗。
* 因此，它提供了一個研究長期暴露於SV40對癌症風險影響的機會

## Materials and Methods

Case-Control Study藉由癌症控制
1. 研究族群：如上述17-30歲的男性。
2. 案件認定條件：腦腫瘤、間皮瘤和非何杰金氏淋巴瘤。
$\Rightarrow$ 我們選擇了所有被 PTF 診斷為上述癌症的男性作為
潛在的案例。 （總計12298）
3. Controls 組: 他們選自接受肺癌或結腸癌治療的病人。（據信兩種癌症與SV40無關。）
$\Rightarrow$ 我們隨機選擇了 5,000 名患有上述癌症的男性。
4. 資料來源：使用以下機構維護的患者治療文件（PTF）退伍軍人管理局。

備註：在這個時代的陸軍人員中，女性比例不到1%。因此將其排除在本研究之外。

## Consider the Race and Age
接下來是關於這個實驗的結果。首先，在不同的癌症類型中，我們考慮種族的影響。由圖表可知：
* 種族分佈存在顯著差異。
* 每種癌症類型患者的入伍年齡相似。
* 所有癌症病例的診斷年齡明顯低於控制。

## Consider Corresponding Adenovirus Vaccination

再來，我們考慮入伍日期的分佈和相應的腺病毒疫苗接種由圖表可知：

腺病毒疫苗暴露率是相似的在case and control中

## Statistical method
> 關於接觸 SV40 污染疫苗的影響

目標：確定接觸SV40污染的疫苗對**腦腫瘤、間皮瘤和非何杰金氏淋巴瘤**風險的影響。
$\Rightarrow$ 我們計算每種癌症的odds ratio。

備註：如果這是可行的，腦癌分析可使用我們在驗證子研究中獲得的組織學診斷來按亞型進行分層。

BTW這裡假設癌症部位之間的癌症判定沒有差異。

## Statistical methods

目標：測試對照組中腺病毒疫苗暴露的盛行率是否反映了基礎人群中的盛行率。
$\Rightarrow$ 我們對每種癌症應用卡方檢定。
$\Rightarrow$ 這比較了以下組別的分佈：對照組和該基地有62萬名軍事人員。

以下是一些細節關於奇數比和所有統計檢定。

## Process of Sample Screening

接著是關於資料篩選的部分：
1. 超過91％的潛在病例和對照成功地與社會安全號碼上的BIRLS進行了匹配(兩個數據庫中的患者姓名完全相同或非常相似。)
2. case（n = 5724）和control（n = 4995）組個別有超過一半的入伍進入了陸軍。
3. 其中482例（8％）潛在病例和251例（5％）對照是在1959年至1961年入伍的。
4. 這些潛在病例（n = 470）和對照（n = 249）中，超過97％在入伍日期時年齡為17-31歲。
5. 約90％在入伍後的10-35年內被診斷出患病。
6. 排除了六例多發性癌症病例和三例對照，最終樣本量為411例癌症病例和221例對照

最後：腦腫瘤181、間皮瘤10和非何杰金氏淋巴瘤220。肺癌107和結腸癌114

## Consider the Race and Age

就上述的資料和統計方法，我們有以下的結果：
首先，在不同的癌症類型中，我們考慮種族的影響：由圖可知，
* 種族分佈存在顯著差異。
* 每種癌症類型患者的入伍年齡相似。
* 所有癌症case組的診斷年齡明顯低於control組。

## Consider Corresponding Adenovirus Vaccination

再來，我們考慮入伍日期的分佈和相應的腺病毒疫苗接種：由圖可知，

* 腺病毒疫苗暴露率在case組和control組中相似控制。

特別是，我們可以看到以下疫苗暴露$OR$和 95% 信賴區間 ($CI$)。

* 腦腫瘤：$OR=0.81$，$CI=[0.52, 1.24]$。
* 間皮瘤：$OR=1.41$，$CI=[0.39, 5.15]$。
* 非何杰金氏症：$OR=0.97$，$CI=[0.65, 1.44]$。

在調整診斷年齡和種族後，這些數據並沒有受到影響。

為了因應可能出現的問題，（1960年5月入伍至1960年7月），該研究進行了排除該組的分析，並得到了類似的結果！

## Discussion

In this case-control study, 我們沒有發現任何證據顯示暴露於腺病毒疫苗與腦腫瘤、間皮瘤或非霍奇金淋巴瘤的風險增加有關。

* 少數病例為間皮瘤和特定腦癌亞型。
$\Rightarrow$ 這限制了我們對這些因發癌症部位的推論。
* 所有腦腫瘤和非何杰金氏淋巴瘤的病例數都很大，相應的比值比 95% 置信區間排除了大的影響。

因此這份paper的結論是，我們的研究提供了反對SV40是這些癌症的原因的證據。

* 這份研究結果與先前關於小兒麻痺疫苗接受者的回顧性cohort study的結果相抵觸。
* 上述實驗室指出人類腫瘤中存在SV40 DNA序列，包括間皮瘤、腦腫瘤和非霍奇金淋巴瘤。

然而，對這些實驗室研究結果的解釋並不簡單。在發現SV40 DNA的腫瘤的多樣性以及這些腫瘤中檢測到的SV40 DNA水平很低的情況下，有可能部分結果是人為的。全球實驗室使用的200多個克隆向量中存在SV40 DNA序列，這可能在實驗室評估期間導致腫瘤組織的污染。許多實驗室在評估腫瘤樣本時，未包括適當的正常組織作為陰性對照，也未在評估過程中採取遮蔽措施。

# Text

Case-Control Study of Cancer among US Army Veterans Exposed to Simian Virus 40-contaminated Adenovirus Vaccine
> 美國陸軍退伍軍人暴露於受猴病毒40污染的腺病毒疫苗之癌症病例對照研究
> Author. Dana E. M. Rollison1, William F. Page, Harriet Crawford, Gloria Gridley, Sholom Wacholder, Jennifer Martin, Richard Miller, and Eric A. Engels.

## Abstract
Simian virus 40 (SV40) was an accidental contaminant of vaccines produced in monkey kidney tissue cultures in the 1950s and early 1960s, including a parenteral adenovirus vaccine given to several hundred thousand US military recruits. Detection of SV40 DNA in tumor tissues by some laboratories suggests that SV40 contributes to human cancers. To determine if entry into US Army service during periods of administration of SV40-contaminated adenovirus vaccine was associated with an increased risk of cancer, the authors conducted a casecontrol study of cancer occurring in male Army veterans who entered service in 1959–1961. Cases of brain tumors (n = 181), mesothelioma (n = 10), and non-Hodgkin’s lymphoma (n = 220) were identified through a Veterans Administration hospital discharge database, as were colon cancer and lung cancer controls (n = 221). Exposure to adenovirus vaccine was assigned on the basis of known periods of adenovirus vaccine administration and dates of Army entry obtained for cancer cases and controls. The odds ratios associated with exposure to SV40-contaminated adenovirus vaccine were 0.81 (95% confidence interval (CI): 0.52, 1.24) for brain tumors, 1.41 (95% CI: 0.39, 5.15) for mesothelioma, and 0.97 (95% CI: 0.65, 1.44) for non-Hodgkin’s lymphoma. These findings do not support a role for SV40 in the development of these cancers.
> 猴病毒40（SV40）是20世紀50年代和60年代初在猴子腎組織培養的疫苗中意外污染的物質，包括給予數十萬名美國軍事新兵的腺病毒疫苗。一些實驗室在腫瘤組織中檢測到SV40 DNA，暗示SV40可能導致人類癌症。==為了確定在SV40污染的腺病毒疫苗施打期間加入美國陸軍是否與癌症風險增加有關==，作者對1959年至1961年入伍的男性陸軍退伍軍人進行了一項癌症病例對照研究。通過退伍軍人事務部醫院出院數據庫識別出腦瘤（n = 181）、間皮瘤（n = 10）和非霍奇金淋巴瘤（n = 220）病例，以及結腸癌和肺癌對照組（n = 221）。根據已知的腺病毒疫苗施打期間和癌症病例及對照組的入伍日期，確定了暴露於腺病毒疫苗的情況。與接觸SV40污染的腺病毒疫苗相關的勝算比分別為腦瘤0.81（95％信賴區間（CI）：0.52，1.24）、間皮瘤1.41（95％CI：0.39，5.15）和非霍奇金淋巴瘤0.97（95％CI：0.65，1.44）。這些結果不支持SV40在這些癌症的發展中扮演角色。

brain neoplasms; lymphoma, non-Hodgkin; mesothelioma; military personnel; simian virus 40; United States Department of Veterans Affairs.

> 腦腫瘤；淋巴瘤，非霍奇金型；間皮瘤；軍事人員；猴病毒40；美國退伍軍人事務部。

##

Simian virus 40 (SV40), a macaque polyomavirus, ==was an accidental contaminant in vaccines produced from monkey kidney tissue cultures in the 1950s and early 1960s.== SV40 was present in the inactivated poliovirus vaccine, which was administered to 10–30 million people, mostly children, in the United States from 1955 to 1961 (1). SV40 was also present in other vaccines, notably adenovirus vaccines given parenterally to several hundred thousand US military recruits (2).

> 猴病毒40（SV40）是一種猕猴多瘤病毒，==在20世紀50年代和60年代初生產的疫苗中是一種意外污染物。== SV40存在於灭活的脊髓灰质炎疫苗中，這種疫苗在1955年至1961年間被施打給美國的1000萬至3000萬人，主要是兒童。SV40也存在於其他疫苗中，特別是給予數十萬名美國軍事新兵的腺病毒疫苗。

These human exposures to SV40 raise concern, because several lines of laboratory evidence suggest that SV40 could play a role in the etiology of some human malignancies SV40 codes for T antigen, a nonstructural protein that can bind to and inhibit **tumor suppressor proteins** (e.g., Mr 53,000 protein (p53) and retinoblastoma protein (pRb)) (see reviews (3, 4)). In experimental rodents, SV40 inoculation leads to the development of various tumors, in particular mesothelioma (5), ependymoma (6, 7), osteosarcoma (8, 9), and leukemia and lymphoma (9). More directly, some laboratory studies have reported the detection of SV40 DNA sequences in a variety of human tumors, including mesothelioma (10, 11), brain tumors (especially ependymoma and choroid plexus tumors) (12), and non-Hodgkin’s lymphoma (13, 14). In contrast, other laboratory studies have not confirmed these findings (15–20).

> 這些人類暴露於SV40的情況引起了關注，==因為幾條實驗室證據線索表明SV40可能在某些人類惡性腫瘤的病因中起作用==。SV40編碼T抗原，==這是一種非結構蛋白，可以結合並抑制**腫瘤抑制蛋白**==（例如，Mr 53,000蛋白（p53）和視網膜母細胞瘤蛋白（pRb））。在實驗性啮齒動物中，==SV40接種導致各種腫瘤的發展，特別是間皮瘤、室管膜瘤、骨肉瘤以及白血病和淋巴瘤。== 更直接地，一些實驗室研究報告在各種人類腫瘤中檢測到SV40的DNA序列，包括**間皮瘤**、**腦腫瘤**（尤其是室管膜瘤和脈絡從腫瘤）以及非霍奇金淋巴瘤。相反，其他實驗室研究未能證實這些發現。

Retrospective cohort studies have not identified increased cancer risk in recipients of SV40-contaminated poliovirus vaccines (see review (21)). A potential limitation of USbased studies has been that only 10–30 percent of lots of poliovirus vaccine in the United States was actually contaminated with live SV40 (1), so assigning SV40 exposure status on the basis of receipt of vaccine could conceivably have prevented researchers from identifying an effect. Additionally, prior retrospective studies of children exposed to SV40-contaminated poliovirus vaccines had little opportunity to detect an increased risk for cancers (such as mesothelioma) that arise primarily in middle- or older-aged adults, given the relatively young ages attained during follow-up.

> 回顧性隊列研究並未發現接受猴病毒40污染脊髓灰質炎疫苗的接種者增加癌症風險（參見文獻回顧）。==美國研究的潛在局限性在於，實際上只有10％至30％的脊髓灰質炎疫苗批次被發現污染有活性猴病毒40，因此僅根據接種疫苗的情況來確定SV40暴露狀態可能會阻礙研究人員識別效應。== 此外，**以往對暴露於猴病毒40污染脊髓灰質炎疫苗的兒童進行的回顧性研究在隨訪期間很少有機會檢測出對於主要發生在中年或老年成人中的癌症（如間皮瘤）風險增加的情況，因為這些研究對象在隨訪期間的年齡相對較年輕。**

Beginning in 1960, the US Army routinely vaccinated recruits against adenovirus to prevent epidemics of acute respiratory illness during basic training (2). As we review below, there is compelling evidence that this vaccine was widely contaminated with SV40. To better understand the potential cancer risk associated with SV40 infection, we conducted a case-control study of cancers occurring in male US Army veterans who entered Army service between January 1959 and December 1961, some of whom received adenovirus vaccine. Cases of brain tumors, mesothelioma, and non-Hodgkin′s lymphoma were identified through a Veterans Administration (later the Department of Veterans Affairs) hospital discharge database, as were a random sample of colon cancer and lung cancer controls. Dates of Army entry were obtained for cancer cases and cancer controls, and exposure to adenovirus vaccine was assigned on the basis of known periods of administration of the vaccine.

> 自1960年起，美**國陸軍開始對新兵進行腺病毒疫苗接種，以防止基礎訓練期間急性呼吸道疾病的流行（參見文獻回顧）**。正如我們下面所述，有令人信服的證據表明，這種疫苗普遍受到猴病毒40的污染。==為了更好地了解與猴病毒40感染相關的潛在癌症風險，我們對進入美國陸軍服役的男性退伍軍人（服役時間為1959年1月至1961年12月）進行了一項癌症病例對照研究，其中一些人接種了腺病毒疫苗。== 通過退伍軍人事務部（後來是退伍軍人事務部）醫院出院數據庫識別出**腦腫瘤**、**間皮瘤**和**非霍奇金淋巴瘤病例**，以及隨機抽樣的結腸癌和肺癌對照組。為了==癌症病例和癌症對照組==獲得了入伍日期，並根據已知的**疫苗施打期間分配了接觸腺病毒疫苗的情況**。

## Materials and Methods

### Production, contamination with SV40, and administration of adenovirus vaccine

> 腺病毒疫苗的生產、SV40污染及接種

In the 1950s, the US Army developed an inactivated vaccine against adenovirus types 4 and 7 (2). In production of the vaccine, adenovirus was propagated in kidney tissue from macaques, many of whom were infected with SV40 (22, 23). This resulted in frequent SV40 contamination of tissue culture systems (22, 23). Furthermore, it is now known that adenoviruses grow extremely poorly in monkey kidney tissue unless SV40 is also present (24). Because SV40 greatly facilitates growth of adenoviruses, it is considered highly likely that, through strong positive selection in the laboratory, nearly all adenovirus seeds and the vaccine pools grown from them in this era quickly became contaminated with SV40 (23, 25). This situation is unlike poliovirus vaccine contamination, which did not occur uniformly, because SV40 is not a necessary cofactor for poliovirus replication in vitro.

> 在1950年代，美國陸軍開發了一種針對腺病毒4型和7型的灭活疫苗。在疫苗的生產過程中，腺病毒在猕猴的腎組織中培養繁殖，而許多猕猴感染了猴病毒40。這導致組織培養系統經常被SV40污染。此外，現在已知腺病毒在猕猴腎組織中的生長非常不良，除非也存在SV40。由於SV40極大地促進了腺病毒的生長，因此人們普遍認為在這個時代，由於實驗室中的強烈正選，幾乎所有的腺病毒種子和從中培養出的疫苗混合物都迅速被SV40污染。這種情況不同於脊髓灰質炎疫苗的污染，因為在體外環境中，SV40對脊髓灰質炎病毒的複製不是必要的輔因子。

In the final step of vaccine manufacture, vaccine pools were treated with formalin to inactivate live adenovirus. However, SV40 is relatively resistant to formalin inactivation. It is unknown how much live SV40 remained in the final adenovirus vaccine product, which would have varied some with manufacturing technique. However, because inactivation techniques were similar for adenovirus and poliovirus vaccines, it is likely that the amount of live SV40 was similar to that which was present, albeit more variably, in the inactivated poliovirus vaccine. Of importance, in 1961, Gerber et al. tested three samples of formalin-inactivated adenovirus vaccine “prepared by [a] domestic manufacturer” (26, p.206), and all samples had live SV40.Furthermore, in a 1956 adenovirus vaccine trial conducted at Fort Dix, New Jersey, 100 percent of evaluated subjects (nine of nine subjects) demonstrated seroconversion to SV40 (23). Thus, the positive selective pressure driving widespread SV40 contamination of adenovirus vaccine pools, the detection of live SV40 in formalin-inactivated vaccine, and the documented seroconversions following vaccination together point to frequent contamination of this vaccine.

> 在疫苗製造的最後步驟中，疫苗混合物被用甲醛處理以灭活活性腺病毒。然而，SV40對甲醛的灭活相對抵抗力較強。目前不清楚最終的腺病毒疫苗產品中還有多少活性的SV40，這在一定程度上取決於製造技術。然而，由於腺病毒和脊髓灰質炎疫苗的灭活技術類似，因此活性的SV40量可能與脊髓灰質炎疫苗中的存在量相似，儘管變化較大。重要的是，在1961年，Gerber等人測試了三個甲醛灭活的腺病毒疫苗樣本，這些樣本是由一家國內製造商製備的，所有樣本都含有活性的SV40。此外，在1956年在新澤西州迪克斯堡（Fort Dix）進行的腺病毒疫苗試驗中，所有評估對象（九名受試者中的九名）均表現出對SV40的血清轉換。因此，推動腺病毒疫苗混合物普遍受到SV40污染的積極選擇性壓力、在甲醛灭活疫苗中檢測到活性SV40以及接種後記錄的血清轉換共同指出這種疫苗的頻繁污染情況。

The US Army began routine use of this parenteral adenovirus vaccine in 1960. Vaccination was administered on the first day of basic training, corresponding closely with the recorded date of entry into service. According to a 1961 article by the chief of the US Army’s Communicable Disease Branch, Office of the US Army Surgeon General, adenovirus vaccine was given universally to all recruits entering Army service from February through April of 1960, but “production difficulties” (2, p. 1126) then interrupted vaccination until August 1960, when vaccination resumed. Recruits were again routinely vaccinated from August 1960 through May 1961, when vaccination ceased because “supplies once more became limited” (2, p. 1126). Other evidence suggests that adenovirus vaccine was withdrawn in 1961 because of concerns regarding contamination with SV40. Specifically, a 1994 history of the Armed Forces Epidemiology Board states that “SV40 was found to be present in the killed adenovirus vaccine as well as in viral seed stocks…. In consequence, the Division of Biologics, National Institutes of Health, acted in 1961 to prevent the distribution of vaccines containing viable SV40 virus” (27,p. 38).

> 美國陸軍於1960年開始常規使用這種腺病毒疫苗。疫苗在基礎訓練的第一天施打，與入伍日期記錄密切相符。根據美國陸軍外科醫生長辦公室傳染病科主任於1961年發表的一篇文章，自1960年2月至4月，所有入伍的新兵均接種了腺病毒疫苗，但之後由於“生產困難”（引自文獻2，第1126頁）的原因中斷，直到1960年8月才恢復接種。從1960年8月至1961年5月，新兵再次接受常規接種，之後由於“供應再次變得有限”（引自文獻2，第1126頁）而停止接種。其他證據表明，由於擔心受到SV40污染，腺病毒疫苗於1961年被撤回。具體來說，根據1994年武裝部隊流行病學委員會的歷史記錄，發現“殺死的腺病毒疫苗以及病毒種子庫中均存在SV40……因此，國家衛生研究院的生物製劑部於1961年采取行動，防止分發含有活性SV40病毒的疫苗”（引自文獻27，第38頁）。

Persons entering the Army in 1959–1961 also had exposures to inactivated poliovirus vaccine possibly contaminated with SV40. For instance, in 1961, 84 percent of US residents aged 15–19 years had previously received one or more doses of inactivated poliovirus vaccine (28). Similar estimates were reported specifically for US military recruits in 1962 (29). Because of concerns regarding the risk of developing poliomyelitis during service, beginning in 1959, the Army required all entering personnel to have received the basic series of three poliovirus vaccinations plus a single booster. However, only a minority of inactivated poliovirus vaccine doses contained live SV40 (1). Thus, the defined periods of uniform exposure to SV40 through adenovirus vaccination still provided an opportunity to study the longterm effects of SV40 exposure on cancer risk.

> 1959年至1961年入伍美國陸軍的人士也有可能接觸到可能受到SV40污染的灭活脊髓灰质炎疫苗。例如，1961年，84%的15至19歲美國居民曾經接種過一劑或多劑的灭活脊髓灰质炎疫苗。具體而言，1962年美國軍方新兵的估計報告也是如此。由於擔心在服役期間患上脊髓灰質炎的風險，自1959年開始，陸軍要求所有入伍人員接種三劑脊髓灰質炎疫苗的基本系列加上一劑單一增強劑。然而，只有少數灭活脊髓灰質炎疫苗含有活性的SV40。因此，通過腺病毒疫苗接種提供了一個研究SV40長期暴露對癌症風險的影響的機會。

### Study population and ascertainment of cases and controls

We conducted a case-control study with cancer controls (30). Cases and controls were restricted to men who entered US Army service between January 1959 and December 1961 at the ages of 17–30 years and who were treated for cancer within the Veterans Administration medical system. The case-defining conditions were brain tumors, mesothelioma, and non-Hodgkin’s lymphoma, each considered as possibly related to SV40. We used as controls patients treated for other cancers (rather than noncancerous conditions), since Veterans Administration referral patterns across cancer types would be somewhat similar. Specifically, controls were selected from patients treated for lung cancer or colon cancer, two types of cancer believed to be unrelated to SV40. Women represented less than 1 percent of Army personnel entering in this era and were excluded from this study

> 我們進行了一項病例對照研究，使用了癌症對照組。研究對象限定在1959年1月至1961年12月入伍的男性，當時年齡在17至30歲之間，並在退伍軍人事務管理系統中接受癌症治療。病例定義為可能與SV40有關的腦腫瘤、間皮瘤和非霍奇金淋巴瘤。我們選擇了接受其他癌症治療（而不是非癌症情況）的患者作為對照組，因為在各種癌症類型之間，退伍軍人事務管理系統的轉診模式可能有些相似。具體而言，對照組是從接受肺癌或結腸癌治療的患者中選擇的，這兩種癌症被認為與SV40無關。在這個時期，女性在入伍的陸軍人員中所佔比例不到1％，因此在本研究中被排除在外。

Cancer diagnoses for both cases and controls were ascertained using the Patient Treatment File (PTF) maintained by the Veterans Administration. The PTF file is a computerized record of hospital discharge diagnoses for all hospital visits covered by the Veterans Administration, and the file provided data for 1969–1996. To ensure uniform PTF ascertainment of cancers in men who entered the Army across the 1959–1961 period, we further required that cancers arise in the period 10–35 years after Army entry (i.e., in 1969–1994 for men entering in 1959 and in 1971–1996 for men entering in 1961).

> 病例和對照組的癌症診斷是通過退伍軍人事務管理局維護的患者治療文件（PTF）確定的。PTF文件是對所有由退伍軍人事務管理局負擔的醫院就診的出院診斷的電腦記錄，該文件提供了1969年至1996年的數據。為確保在1959年至1961年間入伍的男性中對癌症的PTF確定具有一致性，我們進一步要求癌症發生在入伍後10至35年的期間內（即對於1959年入伍的男性為1969年至1994年，對於1961年入伍的男性為1971年至1996年）。

Notably, the PTF does not have information on the branch of service or date of entry into service, so this information was obtained by linking to another database (see “Matching and exposure assessment” below). Thus, we first identified potential cases and controls as individuals with cancer who might have entered the Army during the period of interest. To do this, we restricted consideration to the PTF diagnoses in males born between January 2, 1928, and December 31, 1944, that is, the only range of birth dates that corresponded to possible entry into the Army in 1959–1961 at ages 17–30 years. We selected all such men with PTF diagnoses of brain tumor (International Classification of Diseases (ICD), Revision 8 or 9, code 191), mesothelioma (ICD code 163), and non-Hodgkin’s lymphoma (ICD codes 200 and 202) as potential cases. To maximize feasibility of matching and exposure assessment for potential controls, which were much more common than cases, we randomly selected 5,000 men with colon cancer (ICD code 153) and 5,000 men with lung cancer (ICD code 162).

> 值得注意的是，PTF沒有有關服役分支或入伍日期的信息，因此這些信息是通過與另一個數據庫進行鏈接獲取的（請參見下文的“匹配和暴露評估”）。因此，我們首先將可能在感興趣的期間入伍的癌症個案和對照組識別為個體。為此，我們僅考慮了生於1928年1月2日至1944年12月31日之間的男性的PTF診斷，這是唯一一個對應於1959年至1961年可能在17至30歲入伍的出生日期範圍。我們選擇了所有具有腦腫瘤（國際疾病分類（ICD）修訂8或9，代碼191）、間皮瘤（ICD代碼163）和非霍奇金淋巴瘤（ICD代碼200和202）PTF診斷的男性作為潛在病例。為了最大程度地提高對潛在對照組（比例遠高於病例）的匹配和暴露評估的可行性，我們隨機選擇了5000名結腸癌（ICD代碼153）和5000名肺癌（ICD代碼162）男性。

### Statistical methods

We calculated the proportion of men who entered the Army during a period of adenovirus vaccination for both cancer cases (brain tumors, mesothelioma, non-Hodgkin’s lymphoma) and controls (lung cancer, colon cancer). To determine the effect of exposure to SV40-contaminated vaccine on the risk of brain tumors, mesothelioma, and non-Hodgkin’s lymphoma, we calculated an odds ratio for each cancer site as the ratio of the odds of exposure in the cancer cases to the odds of exposure in the cancer controls (colon cancer and lung cancer combined). Brain cancer analyses were stratified by subtype when available, using histologic diagnoses obtained in our validation substudy. Odds ratios were adjusted for age at diagnosis and race using logistic regression. Under the assumptions that SV40 is not associated with colon cancer or lung cancer and that ascertainment of cancers does not differ across the cancer sites, this approach produces a valid estimate of the effect of SV40 on risk of brain tumors, mesothelioma, and non-Hodgkin’s lymphoma (30). To test whether the prevalence of adenovirus vaccine exposure in controls reflected that in the underlying population, we compared the distribution of dates of Army entry among lung cancer and colon cancer controls with the distribution among the underlying cohort of 620,000 servicemen using a χ2 test.

> 我們計算了在腺病毒疫苗接種期間入伍的男性比例，分別對病例（腦腫瘤、間皮瘤、非霍奇金淋巴瘤）和對照組（肺癌、結腸癌）進行了計算。為了確定接觸到受SV40污染的疫苗對腦腫瘤、間皮瘤和非霍奇金淋巴瘤風險的影響，我們計算了每個癌症部位的勝算比，作為病例中暴露於疫苗的勝算與對照組（結腸癌和肺癌組合）中暴露於疫苗的勝算之比率。在可用時，腦癌分析按亞型進行分層，使用我們在驗證子研究中獲得的組織學診斷。使用 logistic 回歸對勝算比進行了年齡和種族調整。在SV40不與結腸癌或肺癌相關且癌症確定在各癌症部位之間沒有差異的假設下，該方法可產生關於SV40對腦腫瘤、間皮瘤和非霍奇金淋巴瘤風險的有效估計。為了測試對照組中腺病毒疫苗暴露的患病率是否反映了基礎人群中的情況，我們使用χ2檢驗比較了肺癌和結腸癌對照組入伍日期的分佈與62萬軍人基礎群體的分佈。

Confidence intervals were calculated for odds ratios, derived using an exact method when the expected counts were less than five. All statistical tests were two sided, and p values of less than 0.05 were considered statistically significant. Statistical analyses were conducted using SAS version 8.0 software (SAS Institute, Inc., Cary, North Carolina).

> 對於勝算比，當預期計數小於五時，使用準確方法計算了置信區間。所有統計檢驗均為雙尾檢驗，p值小於0.05被認為具有統計學意義。統計分析使用SAS 8.0軟件進行（SAS Institute, Inc.，北卡羅來納州凱里）。

## Results

A total of 12,298 potential cases (diagnoses of brain tumor, mesothelioma, non-Hodgkin’s lymphoma) and 10,000 potential controls (diagnoses of colon cancer, lung cancer) were ascertained through PTF records. Over 91 percent of potential cases and controls were successfully matched to the BIRLS on Social Security number (figure 1). Names of patients were identical or very similar between the two databases. Almost half of the matched potential cases (n = 5,724) and controls (n = 4,995) entered the Army, of whom 482 (8 percent) potential cases and 251 (5 percent) controls entered in 1959–1961. Over 97 percent of these potential cases (n = 470) and controls (n = 249) were 17–31 years of age on the date of entry, and approximately 90 percent were diagnosed in the period 10–35 years after entry. Six cases and three controls with multiple cancers were excluded, yielding a final sample size of 411 cancer cases and 221 controls (figure 1).

> 通過PTF記錄，共確定了12298例可能的病例（診斷為腦腫瘤、間皮瘤、非霍奇金淋巴瘤）和10000例可能的對照（診斷為結腸癌、肺癌）。**超過91％的潛在病例和對照成功地與社會安全號碼上的BIRLS進行了匹配（見圖1）。**==兩個數據庫中的患者姓名完全相同或非常相似。== ==幾乎一半的匹配潛在病例（n = 5724）和對照（n = 4995）入伍進入了陸軍，其中482例（8％）潛在病例和251例（5％）對照是在1959年至1961年入伍的。== 這些潛在病例（n = 470）和對照（n = 249）中，超過97％在入伍日期時年齡為17-31歲，約90％在入伍後的10-35年內被診斷出患病。排除了六例多發性癌症病例和三例對照，最終樣本量為411例癌症病例和221例對照（見圖1）。

There were 181 cases of brain tumors, 10 cases of mesothelioma, and 220 cases of non-Hodgkin’s lymphoma identified. We included as controls 107 colon cancers and 114 lung cancers. There were significant differences in race by cancer type, with higher percentages of Whites among brain tumor, mesothelioma, and non-Hodgkin’s lymphoma cases than among colon cancer and lung cancer controls (table 1). The ages at Army entry were similar for those with each cancer type (table 1). The age at diagnosis was significantly younger for all cancer cases combined than for colon cancer and lung cancer controls (mean age: 46.5 vs. 49.4 years; p < 0.0001).

> 已識別出181例腦瘤、10例間皮瘤和220例非霍奇金淋巴瘤的病例。我們將107例結腸癌和114例肺癌作為對照組。在不同癌症類型之間，種族分布存在顯著差異，腦瘤、間皮瘤和非霍奇金淋巴瘤患者中的白人比結腸癌和肺癌對照組中的白人百分比高（表1）。入伍時的年齡對於每種癌症類型的患者是相似的（表1）。診斷時的年齡對於所有癌症病例而言顯著年輕於結腸癌和肺癌對照組（平均年齡：46.5歲 vs. 49.4歲；p < 0.0001）。

Table 2 presents the distributions of dates of Army entry and corresponding exposure to adenovirus vaccine for cases and controls. The prevalence of adenovirus vaccine exposure was similar in cases and controls. Specifically, the odds ratios for vaccine exposure were 0.81 (95 percent confidence interval (CI): 0.52, 1.24) for brain tumors, 1.41 (95 percent CI: 0.39, 5.15) for mesothelioma, and 0.97 (95 percent CI: 0.65, 1.44) for non-Hodgkin’s lymphoma. These odds ratios were not affected by adjustment for age at diagnosis and race (table 2). Additionally, as shown in table 2, the dates of entry into Army service of controls were similar to those of the underlying cohort of 620,000 men (χ2 4df = 2.03; p = 0.73).

> 表2呈現了入伍日期和相應腺病毒疫苗接種情況的分佈情況，包括病例和對照組。腺病毒疫苗接種的患病率在病例和對照組中是相似的。具體而言，對於腦瘤，腺病毒疫苗接種的勝算比為0.81（95％置信區間（CI）：0.52，1.24），對於間皮瘤為1.41（95％CI：0.39，5.15），對於非霍奇金淋巴瘤為0.97（95％CI：0.65，1.44）。這些勝算比在年齡診斷和種族調整後並未受到影響（表2）。此外，如表2所示，對照組入伍日期與620,000名男性基本隊伍的入伍日期相似（χ2 4df = 2.03; p = 0.73）。

To investigate the possibility that our assignments of vaccine exposures were off by 1 or 2 months, we lagged the cutpoints used to define exposure to adenovirus vaccine by 1 and 2 months and observed no appreciable change in odds ratios (data not shown). To accommodate the possibility that exposure was incorrectly assigned for the relatively small group of men who entered the Army from May 1960 to July 1960, the time period corresponding to the brief interruption in adenovirus vaccination, we conducted analyses excluding this group, and similar results were obtained (data not shown).

> 為了探討我們對疫苗接種的分配是否錯誤了1或2個月的可能性，我們將用於定義對腺病毒疫苗暴露的切點延遲了1和2個月，並觀察到勝算比沒有明顯的變化（未顯示數據）。為了應對可能存在的問題，即1960年5月至1960年7月進入陸軍的男性這一相對較小的群體，這段時間對應於腺病毒疫苗接種的短暫中斷，我們進行了分析，排除了這一群體，結果類似（未顯示數據）。

## Discussion

In our case-control study of cancers occurring among men who entered the US Army in 1959–1961, we found no evidence that exposure to adenovirus vaccine was related to an increased risk for brain tumors, mesothelioma, or non-Hodgkin’s lymphoma. The numbers of cases of mesothelioma and specific brain cancer subtypes were small, limiting our inferences about these sites. However, the numbers of all brain tumors and non-Hodgkin’s lymphoma cases were substantial, and the 95 percent confidence intervals for the corresponding odds ratios excluded an effect of large magnitude (table 2). As we review, substantial evidence indicates that the Army’s adenovirus vaccine was widely contaminated with SV40. We conclude that our study provides evidence against SV40 as a cause of these cancers.

> 在我們針對1959年至1961年入伍美國陸軍男性進行的癌症病例對照研究中，我們並未找到證據表明接觸腺病毒疫苗與腦瘤、間皮瘤或非霍奇金淋巴瘤風險增加有關。間皮瘤和特定腦癌亞型的病例數量較少，限制了我們對這些部位的推論。然而，所有腦瘤和非霍奇金淋巴瘤病例的數量都相當可觀，相應的勝算比95％信賴區間排除了大幅度效應（表2）。正如我們所回顧的，大量證據表明陸軍的腺病毒疫苗被廣泛污染了SV40。我們的研究結論是，我們的研究提供了反對SV40作為這些癌症原因的證據。

This retrospective case-control study is the first investigation of cancer risk associated with receipt of the Army’s early adenovirus vaccine. Our study has two important strengths. First, exposure assignment in our study was based on specific information about when individual subjects entered the Army and published data on the Army’s use of adenovirus vaccine. Given the documented contamination of this vaccine with SV40, we were able to assign SV40 exposure to subjects with reasonable confidence. Second, we captured cancers diagnosed up to 35 years after entry into service, when men had reached an age in their fifties or sixties. At this age, mesothelioma and non-Hodgkin’s lymphoma incidence begin to increase, which facilitated our evaluation of the role of SV40 in these malignancies.

> 這個回顧性病例對照研究是對美軍早期腺病毒疫苗接種與癌症風險相關性的首次調查。我們的研究有兩個重要的優勢。首先，我們的研究中暴露分配是基於有關個別受試者何時入伍以及軍方使用腺病毒疫苗的已發表資料。鑒於該疫苗已被證實受到SV40污染，我們能夠相對有信心地將SV40暴露分配給受試者。第二，我們捕獲了入伍後最多35年內被診斷出的癌症病例，當時這些男性已達到50多歲或60多歲的年齡。在這個年齡段，間皮瘤和非霍奇金淋巴瘤的發病率開始增加，這有助於我們評估SV40在這些惡性腫瘤中的作用。

Several additional points should be considered when interpreting our data. We relied on Veterans Administration records to identify cancer cases and controls, but only an estimated 40 percent of veterans eligible for Veterans Administration care are actually patients in the Veterans Administration system. Although Veterans Administration patients are not representative of all Army veterans, it is unlikely that utilization of the Veterans Administration for medical care or identification in the PTF differed by exposure to adenovirus vaccine, which was defined by five alternating calendar time periods over a very short (i.e., 3-year) interval. Furthermore, we explicitly tested the possibility that ascertainment bias might have affected our results by comparing the distribution of these periods of alternating exposure among colon cancer and lung cancer controls with that of the underlying cohort of 620,000 Army servicemen and observed no appreciable difference. Thus, we do not believe that ascertainment of cancers through the PTF introduced bias.

> 在解釋我們的數據時，應考慮幾個額外的要點。我們依賴退伍軍人事務管理局的記錄來識別癌症病例和對照組，但只有估計的40％有資格接受退伍軍人事務管理局的護理的退伍軍人實際上是該系統的病人。雖然退伍軍人事務管理局的病人並不代表所有陸軍退伍軍人，但利用退伍軍人事務管理局進行醫療保健或在PTF中識別的情況不太可能因腺病毒疫苗的接種情況而有所不同，該情況由五個交替的日曆時間段定義，在一個非常短暫（即3年）的間隔內。此外，我們明確測試了可能存在的確認偏差可能會影響我們的結果，方法是比較這些交替暴露期間在結腸癌和肺癌對照組中的分佈與620,000名陸軍服役人員基本群體的分佈，結果沒有明顯差異。因此，我們認為通過PTF確定癌症的方法並未引入偏差。

Misinformation recorded in the PTF or the BIRLS could have resulted in misclassification of case or control status or exposure, biasing our results toward the null. In our effort to validate cancer diagnoses, we were able to obtain Veterans Administration medical records for only a minority of cases, including only four mesotheliomas. Nonetheless, this substudy confirmed the cancer diagnoses in most subjects with available records. Importantly, the proportion of cancer diagnoses verified by Veterans Administration records did not differ by adenovirus vaccine exposure. Our validation substudy of adenovirus vaccine exposure found that there was minimal error in the recording of Army entry dates in the BIRLS. Additionally, risk estimates based on varying cutpoints of Army entry dates did not indicate systematic errors in the assignment of vaccine exposure.

> PTF或BIRLS中的錯誤信息可能導致病例或對照組狀態或暴露的錯誤分類，使我們的結果偏向無效。在我們努力驗證癌症診斷時，我們僅能獲得少數案例的退伍軍人事務管理局醫療記錄，其中僅包括四例間皮瘤。儘管如此，這個子研究確認了大多數具有可用記錄的受試者的癌症診斷。重要的是，通過退伍軍人事務管理局記錄核實的癌症診斷比例在腺病毒疫苗接種情況下沒有差異。我們對腺病毒疫苗接種情況的驗證子研究發現，在BIRLS中記錄陸軍入伍日期方面幾乎沒有錯誤。此外，基於不同陸軍入伍日期的風險估計並未顯示在疫苗接種情況分配中存在系統性錯誤。

We also considered the possibility that exposure to poliovirus vaccine could have affected our ability to observe an association between adenovirus vaccination and cancer. As noted above, all men in our study were likely inoculated with potentially SV40-contaminated poliovirus vaccine. However, given the Army’s policy regarding poliovirus vaccination, receipt of poliovirus vaccine was uniform across the 1959–1961 period and therefore would not have differentially affected those exposed or unexposed to adenovirus vaccine. Additionally, only a minority of the inactivated poliovirus vaccine doses contained SV40, and not all exposures to SV40-contaminated poliovirus vaccine would have led to infection. Thus, the added exposures to SV40 from the contaminated adenovirus vaccine should have led to observable increases in cancer risk at the hypothesized sites, if SV40 infection truly increases cancer risk.

> 我們還考慮到，接受脊髓灰質炎疫苗可能會影響我們觀察腺病毒疫苗接種與癌症之間關聯性的能力。正如上文所述，我們研究中的所有男性很可能都接種了潛在受到SV40污染的脊髓灰質炎疫苗。然而，考慮到陸軍對脊髓灰質炎疫苗接種的政策，在1959年至1961年期間接受脊髓灰質炎疫苗的方式是一致的，因此不會對接受或未接受腺病毒疫苗的人產生不同影響。此外，只有少數的灭活脊髓灰質炎疫苗含有SV40，且並非所有接觸到受到SV40污染的脊髓灰質炎疫苗的人都會感染。因此，來自受污染的腺病毒疫苗的SV40的額外暴露應該會在假設的部位引起可觀的癌症風險增加，如果SV40感染真的會增加癌症風險的話。

Our negative results and those from prior retrospective cohort studies of poliovirus vaccine recipients conflict with laboratory reports of the presence of SV40 DNA sequences in human tumors, including mesothelioma, brain tumors, and non-Hodgkin’s lymphoma. Nonetheless, interpretation of the results of these laboratory studies is not straightforward. The diversity of tumors in which SV40 DNA has been found and the low levels of SV40 DNA that are detected in these tumors raise the possibility that some findings could be artifactual. SV40 DNA sequences are present in over 200 cloning vectors used by laboratories worldwide, which might conceivably lead to contamination of tumor tissues during laboratory evaluation. Many laboratories have not included appropriate normal tissues as negative controls and have not incorporated masking in their evaluation of tumor specimens. In other studies, SV40 DNA was not detected in human tumors or was detected infrequently, and SV40 DNA sequences were not reproducibly identified in a multilaboratory study in which mesothelioma specimens were evaluated blindly. Finally, case-control studies utilizing serologic assays for SV40 infection have identified only low levels of SV40 antibody reactivity, suggesting that SV40 is an uncommon human infection, and have failed to detect an association between SV40 serostatus and cancer.

> 我們的研究結果與先前關於脊髓灰質炎疫苗接受者的回顧性队列研究的結果相抵觸，這些結果與實驗室報告相矛盾，後者指出人類腫瘤中存在SV40 DNA序列，包括間皮瘤、腦腫瘤和非霍奇金淋巴瘤。然而，對這些實驗室研究結果的解釋並不簡單。在發現SV40 DNA的腫瘤的多樣性以及這些腫瘤中檢測到的SV40 DNA水平很低的情況下，有可能部分結果是人為的。全球實驗室使用的200多個克隆向量中存在SV40 DNA序列，這可能在實驗室評估期間導致腫瘤組織的污染。許多實驗室在評估腫瘤樣本時，未包括適當的正常組織作為陰性對照，也未在評估過程中採取遮蔽措施。在其他研究中，未在人類腫瘤中檢測到SV40 DNA，或者僅偶爾檢測到，而在一項多實驗室研究中，在盲檢評估間皮瘤樣本時，未能重現識別出SV40 DNA序列。最後，利用血清學檢測法進行的病例對照研究只識別出少量的SV40抗體反應，這表明SV40是一種罕見的人類感染，並未能檢測到SV40血清狀態與癌症之間的關聯。

In summary, we did not find an association between exposure to SV40-contaminated adenovirus vaccine among Army service personnel and risk of brain tumors, mesothelioma, or non-Hodgkin’s lymphoma. With the recent development of sensitive and specific serologic assays for SV40 infection, future epidemiologic studies will need to reexamine the possible association between SV40 infection and cancer. Laboratory-based investigations utilizing molecular and serologic techniques must be based on rigorous study designs, incorporating the inclusion of appropriate controls and masking. While our results point away from SV40 as a cause of cancer, additional useful information may be gleaned from such studies.

> 總的來說，我們未發現陸軍人員接觸受SV40污染的腺病毒疫苗與腦腫瘤、間皮瘤或非霍奇金淋巴瘤風險之間存在關聯。隨著對SV40感染進行敏感和特異性血清學檢測方法的最新發展，未來的流行病學研究將需要重新檢視SV40感染與癌症之間可能的關聯。利用分子和血清學技術進行的實驗室研究必須建立在嚴格的研究設計基礎上，包括適當控制組的納入和遮蔽。雖然我們的結果排除了SV40作為癌症原因的可能性，但這些研究可能提供了其他有用的信息。

# Note

inactivated poliovirus vaccine 滅活小兒麻痺疫苗
polio vaccine 小兒麻痺疫苗