February 13, 2007 — 一项新的
检验可以区分肠胃道基质
肿瘤（GIST）与子宫肉瘤（LMS），两种几乎相同的
癌症，但两种癌症的治疗疗程不同；这项新的检验仅需根据两种基因，但比目前利用免疫染色的方式准确。
该检验是根据休士顿德州大学安德森癌症中心与华盛顿西雅图系统生物机构合作研发，描述这项试验的文献线上发表于2月13日的美国国家科学院院刊(The Proceedings of the National Academy of Sciences)期刊中。
安德森肌肉瘤中心Jonathan Trent医师在一项访谈中向Medscape表示，我们希望这项检验可以在6至12个月之内上市；同时，医师们可以将样本寄给我们进行检验。
【诊断错误可能代表生存或是死亡】
Trent医师表示，虽然这些都是罕见的肿瘤，诊断错误可能会是很严重的，这可能代表生与死之间的差异，因为这两种疾病的治疗方式不同；GIST与LMS是很相似的肿瘤，这两种肿瘤都是来自肠胃道的平滑肌细胞，且可能在显微镜下不易分辨。
Trent医师表示，这两种肿瘤的治疗方式迥异，GIST对选择性酪氨酸抑
制剂imatinib（Gleevec，由诺华药厂制造），且对约80%的病例有效，然而LMS以化学疗法治疗，约有60%以上的反应率；但是令一方面，GIST对于化学疗法一点反应都没有，且LMS对于imatinib没有反应，因此，诊断错误带来的问题可能会是很严重的；如果你将罹患LMS病患误诊为GIST，且以化学疗法治疗该名病患，这将无法使这些肿瘤缩小，可能造成毒性且没有好处。
研究者在这项文献中表示，目前，肿瘤之间的差异是利用KIT免疫染色分辨，但这是主观的，且因细胞异源性而可能造成伪阴性的诊断；准确率大约87%，但不确定的病例需要进一步且耗费时间的分析。
Trent医师表示，KIT免疫染色在分辨这两种肿瘤上效果并不好，这是非常主观的，且这两种肿瘤都会表现KIT，造成伪阳性结果。
Trent医师的评论是，这个新检验是项进步，它可以定量而非主观的判定，且目前为止看起来是100%准确；这项检验根据两种基因，如果OBSCN，其RNA表现比C9orf65高，则该肿瘤为GIST，如果反过来，则该肿瘤为LMS。
Trent医师表示，这是项非常崭新的方法，该团队使用一种称为尖端记分配对的分析技术来分析成对基因，而非其他系统一般利用的多重基因方式；我们寻找41,000基因的各式组合，且希望范围缩小到4至8个基因，但令人惊讶的，我们发现仅需要2种基因。
当研究者针对68个高度分化GIST与LMS样本利用微阵列方式分析，该检验成功地区分出67个样本（准确率97.8%）；研究者表示，唯一例外的是，其中一个样本此两种基因的表现刚好各占一半。
他们针对20个样本重复进行（包括基因表现各占一半的样本），但是这次使用更准确的测量方式，测试结果够准确确认出20个样本（准确率100%）。
除此之外，研究者进一步根据19个独立肿瘤且未包括在微阵列分析的样本进行分析；研究者评论，在他们报告其检验正确地确认所有19个肿瘤样本之前（14个为GIST，5个为LMS），这是个很重要的检验。
这两个基因分类系统至今在我们所选出的样本测试中尚未失败过，当然，这不代表我们可以预期这项分类系统可以完美地应用于所有未来的病例，但是，根据我们目前为止所收集到的证据，对于这项检验的高准确性仍有很大的期待。
Simple Test Differentiates GIST From Leiomyosarcoma
By Zosia Chustecka
Medscape Medical News
February 13, 2007 — A simple new test can differentiate between gastrointestinal stromal tumor (GIST) and leiomyosarcoma (LMS), 2 nearly identical cancers that require radically different courses of treatment. The new test is based on just 2 genes but is more accurate than currently used tests involving immunostaining.
Developed by researchers at the University of Texas MD Anderson Cancer Center, in Houston, working in collaboration with the Institute for Systems Biology, in Seattle, Washington, the test is described in a paper published online February 13 in the Proceedings of the National Academy of Sciences.
"We hope that the test will be commercially available in about 6 to 12 months," Jonathan Trent, MD, PhD, from sarcoma center at MD Andersen, told Medscape in an interview. "In the meantime, clinicians can send samples to us for testing."
Misdiagnosis Can Mean Difference Between Life and Death
Although these are rare tumors, the implications for a misdiagnosis can be grave, Dr. Trent commented: "It can mean the mean the difference between life and death, because the diseases are treated so drastically differently." GIST and LMS are very similar tumors, both originating in the smooth-muscle cells of the gastrointestinal tract, and can be indistinguishable microscopically. But they are treated in radically different ways, Dr. Trent continued. GIST responds very well to treatment with the selective tyrosine kinase inhibitor imatinib (Gleevec, Novartis), with responses in up to 80% of cases, whereas LMS is treated with chemotherapy, with responses in up to 60% of cases or more. "But the opposite is not true — GIST does not respond at all to chemotherapy and LMS does not respond at all to imatinib." Hence, the consequences of a misdiagnosis can be devastating: "If you misdiagnosed someone with GIST as having LMS and treated them with chemotherapy, it almost certainly would not work to shrink their tumor, and it almost certainly would result in significant toxicity without any benefit."
At present, differentiation between the tumors is made by KIT immunostaining, but this is subjective and variable due to cellular heterogeneity that may result in false-negative diagnoses, the researchers comment in their paper. It is about 87% accurate, but uncertain cases require further intensive and time-consuming additional analyses.
"KIT immunostaining is not a great test to distinguish the 2," Dr. Trent commented. "It's very subjective, and KIT can be expressed by both types of tumors, leading to false-positive results."
"The new test is an improvement, it's quantitative rather than subjective, and so far it looks to be close to 100% accurate," Dr. Trent commented. It is based on 2 genes. If the gene OBSCN expresses more of its RNA than the gene C9orf65, the tumor is GIST. If it's the other way around, the tumor is LMS.
"This is very novel methodology," Dr. Trent commented. The team used a process knows as top-scoring-pair analysis to look at pairs of genes, instead of the more common multiple-gene approach that has been used in other systems. "We looked at 41,000 genes in all possible combinations, and we were hoping to narrow it down to about 4 to 8 genes that were relevant — but, amazingly, we found that we needed only 2 genes."
When the researches applied the test to microarray data of 68 well-characterized GIST and LMS samples, the test successfully identified 67 of the 68 samples (97.8% accuracy). The exception was 1 tumor that showed a nearly 50-50 split between the 2 genes, the researchers report.
They repeated the trial in 20 of the microarray samples (including the sample that showed a near-equal split in gene expression) but this time used a more accurate measurement procedure on the 2 genes: this time the test correctly identified 20 of the 20 samples (100% accuracy).
In addition, the researchers carried out a further trial in 19 additional independent tumor samples that had not been included in the microarray data. "This is the most important test of our classifier," the researchers comment, before reporting that their test correctly identified all 19 tumors (14 GIST and 5 LMS).
"Thus, the 2-gene classifier chosen has yet to fail on any sample we have chosen," the researchers comment. "Of course, this does not mean that we can expect the classifier to perform perfectly on all future cases, but based on the evidence accumulated to date, there is a strong expectation of high accuracy."
Proc Natl Acad Sci. Published online February 13, 2007.
作者：
Zosia Chustecka 2007-6-20