【云栈自译，欢迎批评。请勿转载，谢谢合作】

肺发育和肺癌间存在遗传平行

基因活动模式为肿瘤分类提供了一条新途径

【DavidShi译文】100多年以来，生物学家们一直这样猜测：癌症生长和胚胎发育有一些共同特征。现在波士顿儿童医院的研究人员为该观点提供了确凿证据。证据显示，通过基因芯片分析和生物信息学技术发现，在早期胚胎肺发育中差异表达的（表达增强或减弱）很多基因也在肺癌中差异表达。

更重要的是，它们显示基因表达分析能预报肺癌的预后，具有和肺发育最早期基因表达模式相似的癌症是所有预后中最为糟糕的。

本文作者之一，医学博士、哲学博士、CHIP 项目主任艾萨克·克海说：“这证实了我们早先的发现，即在理解癌症时正常器官发育的重要性，我们的观察可能转化成更为精确的预后，并帮助我们针对治疗应用，以识别癌症生长机制。”

肺癌是全球癌症致死中病例中的主要病因，有多种亚型，但常被错误地分类，延误了恰当的治疗。此外，在同一亚型的不同癌症在侵袭性上可能有变化。

为了找到肺癌分类的更好方法，刘红叶（音）博士和儿童医院信息学项目(CHIP)里的同事们一起，搜集了186位病人的肿瘤样品，检查了肿瘤的基因活动，并和大鼠正常胚胎肺发育中的基因活动相比较。他们同样地检查了17份正常肺组织样品。从已知的人鼠同源的3500个基因中，他们检测出596个基因：在肺癌和肺发育中，这596个基因的活动都发生了变化。

通过将肺发育的固定轨迹（natural trajectory）作为一个框架，刘和同事们能够预知腺瘤病人的成活情况（腺瘤是肺癌最常见的类型，并且是他们唯一有存活数据的类型）。基因表达模式和极早期肺发育最相象的肿瘤的预后最差，而基因表达模式和晚期肺发育相象的肿瘤的预后最好。即使在单一腺瘤里，亚型、I期疾病和存活时间都随基因表达模式而变化。正常肺组织基因表达模式和肺晚期发育中的表达模式类似。

“此前，癌症和器官发育相关的观点没有被确证，或者说没有被显著性差异的数据证明，”刘说。“发育预期给了我们一种理解癌症的新机制”。

研究人员还发现，良性肿瘤这一种肺癌亚型的基因表达模式与其他类型显著不同。在组织切片检查时，良性肿瘤看起来和小细胞肺癌非常相象，二者常相混淆，而他们的预期寿命和最佳治疗方法却变化很大。“通过分子分析，我们能区别这两种癌症，”刘说。

此外，通过集中研究癌－发育之间最相关的100种基因，刘和同事们发现与肺癌发育关键性生物通路有关的三组基因，其中一些基因可能成为潜在的药物作用目标。个别基因具有干细胞样的特征。

刘的这项工作是根据2004年克海和阿尔文?克博士（阿尔文?克博士是本研究中刘的另一位同事）的一项研究展开的。2004年的研究发现，成神经管细胞瘤（medulloblastoma）这种小儿脑瘤和发育最早期的小脑有很多共同的遗传学特征。(www.childrenshospital.org/newsroom/Site1339/mainpageS1339P1sublevel81.html)。

--------------------------------------------------------------------------------

在腺癌中活动加剧的基因（绿圈）在肺发育早期趋于活跃，而活动降低的基因（紫圈）在发育晚期变得活跃起来。

--------------------------------------------------------------------------------

【下面为英文原文】

Genetic parallels found between lung development and lung cancer

Gene activity patterns provide a new way to classify tumors

For over 100 years, biologists have speculated that cancer growth shares common features with embryonic development. Researchers at Children's Hospital Boston now provide solid evidence for this idea, showing through gene-chip analyses and bioinformatics techniques that many genes that are differentially expressed (turned "up" or "down") during early embryonic lung development are also differentially expressed in lung cancer.

More importantly, they show that gene-expression profiling can predict a lung cancer's prognosis, and that cancers whose gene expression pattern resembles gene expression during the earliest stages of lung development have the worst prognosis of all.

"This confirms our earlier finding of the importance of normal organ development in understanding cancer," says Isaac Kohane, MD, PhD, director of the CHIP program and a co-author on the paper. "Our observations might translate into more accurate prognoses and help us identify mechanisms of cancer growth that can be therapeutically targeted."

Lung cancer, the world's leading cause of cancer deaths, has many known subtypes, but it is commonly misclassified, delaying appropriate treatment. In addition, cancers within a subtype may vary in their aggressiveness.

Seeking a better way to classify lung cancers, Hongye Liu, PhD, and colleagues in the Children's Hospital Informatics Program (CHIP) examined gene activity in tumors from 186 patients and compared it with the gene activity that occurs during normal embryonic lung development in mice. They also examined 17 samples of normal lung tissue. Starting with 3,500 genes known to be common to mice and humans, they identified 596 genes whose activity was altered both in lung tumors and during lung development.

Using the natural trajectory of lung development as a framework, Liu and colleagues were able to predict survival in patients with adenocarcinoma (the most common type of lung cancer, and the only type for which they had survival data). Tumors with gene expression patterns most like those during very early lung development had the worst prognosis, while tumors with gene expression patterns resembling those seen late in lung development had the best prognosis. Even within a single adenocarcinoma subtype ?stage I disease ?survival times varied according to gene expression patterns. Gene expression patterns in normal lung tissue resembled those seen in late in lung development.

"Before, the idea that cancer and organ development are related was not quantified or statistically significantly demonstrated," says Liu. "The development perspective gives us a new mechanism for understanding cancer."

The researchers also found that one lung cancer subtype, carcinoid tumors, have a gene expression profile distinct from all the others. When biopsy specimens are examined, carcinoid looks very similar to small-cell lung cancer, and the two are often mistaken for each other, yet their life expectancy and optimal treatments are very different. "By molecular profiling, we can distinguish these two cancers," Liu says.

In addition, focusing on the 100 genes with the greatest cancer/development correlation, Liu and colleagues found three groups of genes that are involved in biological pathways believed to be key in lung cancer development, and some of the genes showed potential as drug targets. Several genes had stem-cell-like characteristics.

Liu's work builds on a 2004 study, in which Kohane and Alvin Kho, PhD (another co-investigator on Liu's study) showed that a pediatric brain tumor called medulloblastoma shares many common genetic features with the cerebellum in its earliest stages of development (www.childrenshospital.org/newsroom/Site1339/mainpageS1339P1sublevel81.html).

--------------------------------------------------------------------------------

Genes whose activity is increased in adenocarcinoma (green circles) tend to be active early in lung development, while genes with reduced activity (magenta) tend to be active late in development.

--------------------------------------------------------------------------------