点击显示 收起
根据研究者在11月14日线上发表于Lancet肿瘤学的研究结果，对于雌性激素受体阴性乳癌患者而言，细胞株的基因表现可能可以预测对化学治疗的反应；虽然需要长期的后续追踪来确认疗程特异性的基因图谱是否可以预测长期预后，但研究结果显示，疗程特异性基因图谱也可以预测接受适当治疗疗程病患的完全病理反应。
来自法国Bordeaux大学Bergonie机构的资深作者Herve Bonnefoi医师向Medscape表示，这项研究是新观念一个绝佳的例子─个人化医学或是治疗；如果这项研究中所看到的阳性预测值与阴性预测值在我们正在进行的确效研究中获得确认，这对于我们治疗乳癌的方式确实是项进步。
他附带表示，这项研究的结果是令人瞩目的，这也是为什么我们必须在很短的时间内确认这项研究发现；在我们的确认研究完成之前，我不会建议使用这些基因图谱。
过去已经有一些单一组别手术前化学治疗试验，对肿瘤组织样本的基因图谱进行研究，大部份的研究结果显示基因图谱可以预测对治疗的临床或是病理学反应；然而，作者表示，这些研究有两个弱点，因此限制了这项研究在常规执业上的用途。
过去并没有研究企图预测随机分派病患接受不同疗程的反应，他们写到，这使得基因图谱用于特定疗程的专一性没有受到证实；另一项弱点是，许多研究收纳不同类型的病患，包括雌性激素受体阳性与阴性病患；雌性激素受体阳性肿瘤已经被证实对于手术前化学治疗的临床与病理反应较差，且这可能导致雌性激素受体状态相关基因与治疗反应基因之间的影响因子。
目前这项研究针对三项前瞻性第三期临床试验，包括European Organization for Research and Treatment of Cancer（EORTC）与10994/Breast International Group（BIG）研究，且比较罹患雌性激素受体阴性妇女接受两种不同手术前化学治疗疗程的反应；这些化学疗程包括：fluorouracil、epirubicin与cyclophosphamide（FEC）投予六个周期；以及taxane类药物docetaxel投予三个周期，之后使用三个周期的epirubicin与docetaxel（TET）。
Bonnefoi医师与其同事分析来自FEC组的66个肿瘤样本、以及TET组的59个肿瘤样本，她们透过确认预测细胞株对于fluorouracil、cyclophosphamide、docetaxel与epirubicin当做同一个药物的反应，且接着合并单一药物基因图谱形成疗程专一的基因图谱；接着计算两个治疗组的疗程专一基因图谱来预测完全病理反应的或然率。
他们发现，基因图谱显著地预测接受适当药物疗程病患的反应；FEC预测因子显示敏感度为96%（28位病患中有27位有效），专一性为66%（38位病患中有25位）；这组的阳性预测值（PPV）为68%（40位病患中有27位）；阴性预测值为（NPV）为96%（26位病患中有25位）；在第二个治疗组中的病患也有类似的情况，TET预测因子的敏感度为93%（27位病患中有25位）、专一性为69%（32位病患中有22位）；PPV为71%（35位病患中有25位）、NPV为92%（24位病患中有22位）。
如果基因图谱被应用到所有病患以决定最适当的治疗，估计的理论病理完全反应率将会是65~70%，这比FEC组与TET组在临床研究中所观察到病理完全反应率的42%与46%要高得多。
研究作者Richard D. Iggo医师表示，虽然这项研究仅收纳雌性激素受体阴性乳癌病患，要确认雌性激素受体阳性病患的基因图表同样是可能的；Iggo是苏格兰伐夫郡圣安卓大学的一位分子医学教授。
他表示，但是其基因图谱可能是不同的；事实上，雌性激素受体阳性病患对于化学治疗反应较差是进行这类研究最主要的障碍，因为要找到足够的病患是极端困难的。
研究者的结论是，疗程特定的基因图谱可能改善治疗成功率，与降低暴露于taxanes类药物不良反应病患的人数。
然而，目前而言，lggo医师同意必须要再进一步确认这些结果之后，才能在临床研究之外使用基因图谱；北卡罗莱纳州杜汉杜克大学的统计学家们所研发出来的预测流程是非常复杂的，且这限制了临床医师在这个时候应用这个流程的可能性。
这项研究由欧盟第六架构计划活性p53资金、Widmer基金会、Medic基金会与Oncosuisse、瑞士国家基金会NNCR分子生物肿瘤学、EORTC转录研究基金会、瑞典癌症学会、King Gustav the Fifth Jubilee基金会、瑞典研究局、美国国家卫生研究院、美国癌症研究协会（AACR）与研究研究第五基金会赞助。
Gene-Expression Signatures Predict Treatment Response in Breast Cancer Patients
By Roxanne Nelson
Medscape Medical News
In breast cancer patients with estrogen-receptor-negative breast tumors, gene-expression signatures based on cell lines might be able to predict response to chemotherapy, researchers report in a study published in the November 14 online issue of the Lancet Oncology. Although long-term follow-up is needed to ascertain whether regimen-specific genomic signatures can also predict long-term survival, study results show that regimen-specific signatures significantly predicted a complete pathological response in patients treated with the appropriate regimen.
"This trial is a very good example of a modern concept — individualized medicine or treatment," senior author Herve Bonnefoi, MD, from the department of medical oncology, Institut Bergonie, University of Bordeaux, France, told Medscape Oncology. "It [will be] a real improvement in the way we treat breast cancer if the positive predictive values and negative predictive values seen in this trial are confirmed in our ongoing validation study."
The results of the study were striking, he added. "That is why we need to confirm it — or not — urgently. I won't recommend use of these signatures before our confirmatory trial."
There have been several single-arm neoadjuvant chemotherapy trials that have reported gene-expression signatures from samples of tumor tissue, with the majority describing signatures that predict clinical or pathological response to treatment. However, the authors note that these trials contain 2 major weaknesses that have restricted their usefulness in routine clinical practice.
None of the previous trials attempted to predict the response of patients randomized to different regimens, they write, which makes the specificity of the signatures for particular regimens unproven. The second weakness is that many of the trials included a mixed cohort of patients with estrogen-receptor-positive tumors and patients with estrogen-receptor-negative tumors. Estrogen-receptor-positive tumors have been shown to have a lower rate of clinical and pathological response to neoadjuvant chemotherapy, and this can lead to a confounding of treatment-response genes with genes linked to estrogen-receptor status.
The current study was conducted in the context of the prospective phase 3 European Organization for Research and Treatment of Cancer (EORTC) 10994/Breast International Group (BIG) trial, and compared 2 neoadjuvant regimens in women with estrogen-receptor-negative breast cancer: fluorouracil, epirubicin, and cyclophosphamide (FEC) for 6 cycles; and the taxane docetaxel for 3 cycles followed by epirubicin plus docetaxel (TET) for 3 cycles.
Dr. Bonnefoi and colleagues analyzed tissue samples obtained from 66 tumors in the FEC group and 59 tumors in the TET group. They constructed response predictors by identifying the genes that predict the response of cell lines to fluorouracil, cyclophosphamide, docetaxel, and epirubicin given as single drugs, and then combined the single-drug signatures to form regimen-specific genomic signatures. The predicted probability of a complete pathological response was then calculated with the regimen-specific genomic signatures in the 2 treatment groups.
They found that the signatures significantly predicted response in patients who received the appropriate drugs. The FEC predictor showed a sensitivity of 96% (27 of 28 patients) and a specificity of 66% (25 of 38 patients). The positive predictive value (PPV) in this group was 68% (27 of 40 patients); the negative predictive value (NPV) was 96% (25 of 26 patients). The results were similar among patients in the second treatment group. The TET predictor had a sensitivity of 93% (25 of 27 patients) and specificity of 69% (22 of 32 patients). There was a PPV of 71% (25 of 35 patients) and a NPV of 92% (22 of 24 patients).
If genomic signatures had been used in all patients to determine the most appropriate therapy, the estimated hypothetical pathological complete response rate would have been 65% to 70%, which is well above the 42% and 46% pathological complete response rates that were observed in the FEC and TET groups in the clinical trial.
Although this study only included women with estrogen-receptor-negative tumors, it might also be possible to determine genomic signatures in patients with estrogen-receptor-positive tumors, explained study author Richard D. Iggo MD, PhD, a professor of molecular medicine at the University of St. Andrews in Fife, Scotland.
"But they may not be the same signatures," he said. "In fact, the low response rate to chemotherapy in the estrogen-receptor-positive group is a major obstacle to doing any kind of definitive study because it is extremely difficult to find enough patients."
The researchers concluded that the clinical application of regimen-specific genomic signatures might improve the success rate of treatment and reduce the number of patients exposed to the adverse effects of taxanes.
Currently, however, Dr. Iggo agrees that it is essential to confirm the results before using the signatures outside a clinical trial. The prediction algorithm was devised by statisticians at Duke University in Durham, North Carolina, and is extremely complex, he added, which eliminates the possibility of clinicians applying it at the moment.
The study was funded by the European Commission Sixth Framework Programme Active p53 grant, Fondation Widmer, Fondation Medic, Oncosuisse, Swiss National Science Foundation NCCR Molecular Oncology, EORTC Translational Research Fund, Swedish Cancer Society, King Gustav the Fifth Jubilee Fund, Swedish Research Council, US National Institute for Health, the American Association for Cancer Research (AACR), and the V Foundation for Cancer Research.
Lancet Oncol. Published online November 14, 2007.