Gene Profiles As Biomarkers Of Prognosis In Cancer Patients

Despite the problems associated with gene profiling approaches that were noted above, there have recently been significant advances in using gene profiling to predict prognosis in cancer patients. For example, van' t Veer and colleagues isolated at diagnosis 78 core samples from breast cancer patients under the age of 55 [36]. Of these patients, all were lymph node-free at the time of diagnosis. By measuring patient outcome over five years (regardless of treatment regimen), it was observed that 34 patients developed distant metastases, whereas 44 patients did not. Microarrays containing 25,000 human genes were prepared by the investigators, to which labeled cDNA samples from the patients and a pooled reference RNA sample were hybridized.

Following supervised data analysis and correct classification by a leave-one-out cross-validation approach, a classifier of 70 genes was developed in which disease outcome could be predicted. When used within the 78-patient sample training set, the genetic profile predicted patient outcome with an accuracy of 83%. In support of the classifier, the 70-gene set predicted with 89% accuracy the prognosis of an additional 19 patients who were not included within the training set.

The utility of the van't Veer genes as prognostic biomarkers has been tested further using a larger study, which included 295 patients [13]. Sixty-one patient samples (from the study in which the 70-gene classifier was developed) were included as a control. Of the patients used, 151 were lymph node-negative and 144 were lymph node-positive. Kaplan-Meier analysis was performed to determine the probability of patients remaining metastatis-free. The 70-gene classifier showed excellent accuracy at predicting prognosis in all patients (p < 0.001), including lymph node-negative patients (which generally have better outcomes). These findings further suggested that the acquisition of metastases in patients is already defined through gene expression in their tumors rather than being acquired late during tumorgenesis. The findings from the two studies clearly indicate that the 70-gene classifier can reliably predict prognosis in breast cancer patients and therefore may be useful in preventing overtreatment in some patients while identifying those that would probably benefit from adjuvant therapy [13]. The 70-gene set (also known as MammaPrint) has recently been approved for use in the United States by the Food and Drug Administration (FDA). The utility of the van't Veer classifier in predicting prognosis for cancers of tissue origins other than breast is currently unknown.

A recent investigation attempted to predict outcome in breast cancer patients using RNA from tumors of 162 patients [11] and from several cell lines to serve as the reference RNA (SW872, WM115, NTERA2, MCF-7, HEPG2, MOLT4, Hs578t, HL60, OVCAR3, C0L0205, and RPMI 8226 cells). Both labeled reference and tumor RNAs were hybridized to arrays consisting of 10,368 unique genes. Analysis of the hybridized arrays by SAM, prediction analysis for microarrays (PAM), and the approaches described in the van't Veer et al. study [36] resulted in the identification of 49 genes that correlated with patient outcome. The gene list, when further reduced to 21 genes by including only those genes observed in all three analyses, displayed 69% accuracy by the leave-one-out cross-validation strategy. Using the gene set, the classifier appeared to predict patient outcome with 65% accuracy using the van 't Veer data and 62% accuracy using data from another study [37] , Considering chance alone can yield as much as 60% accuracy; the gene set above was not validated. Furthermore, the expression of the van't Veer prognostic genes was unable to classify the patients successfully in terms of disease outcome, though over half of the genes identified in the van't Veer study were not present on the microarrays used. This underscores the difficulties in comparing across microarray experiments when different array platforms are used.

In addition, successful prediction of prognosis using this subset of the 70-gene classifier may have been increased by dividing patients according to lymph node status as was done in the van't Veer follow-up study [13].

While the van't Veer classifier shows significant promise, it has yet to be used widely in a clinical setting. An alternative gene classifier has undergone considerable validation in a clinical setting. This classifier was developed by selecting 250 candidate genes that showed promise as prognostic biomarkers, based on literature searches and previous microarray experiments [13,19,28,38]. The expression of these genes was then measured in tumor biopsies of 447 patients associated with three independent clinical trials. Analysis of the relationship between the tumor expression of each gene and patient outcome resulted in the identification of 21 prognostic genes, including 16 cancer-related genes and five reference genes. This 21- gene prognostic classifier is now known as the Oncotype DX profile [15]. Interestingly, five of the 16 genes in the Oncotype DX profile were identified previously in the study by van 't Veer et al. [36]. The Oncotype DX genetic profile measures the probability of disease recurrence in patients diagnosed with estrogen receptor-positive, lymph node-negative breast cancer who were treated with tamoxifen. The likelihood of recurrence is measured using a recurrence score, which is calculated by measuring the expression of the 21 genes and converts these values into a score ranging from 1 to 100. Low risk of recurrence is determined by a score of <18, while ranges from 18 to 30 and >31 indicates intermediate and high risk of recurrence, respectively. The Oncotype DX gene set has been validated in three additional clinical trials [15,39,40] and is currently undergoing further validation in a phase III study known as TAILORx. Furthermore, the Oncotype DX gene set has been studied further for its ability to predict factors other than disease recurrence. The recurrence score was recently used to measure recurrence risk in 651 patients who were randomly assigned to treatment with tamoxifen or tamoxifen and chemotherapy [41] . The 10-year distant recurrence rates were determined for the two treatment groups and it was revealed that low- and intermediate-risk patients (as predicted by RS) did not benefit from chemotherapy. In contrast, high-risk patients derived great benefit from chemotherapy. Therefore, the Oncotype DX recurrence score appears capable of identifying in a group of patients who typically have a relatively low risk of disease recurrence those who will probably benefit from chemotherapy. This, in turn, would enable us to spend health care resources more efficiently by providing chemotherapy only to those who will benefit from treatment.

10 Ways To Fight Off Cancer

10 Ways To Fight Off Cancer

Learning About 10 Ways Fight Off Cancer Can Have Amazing Benefits For Your Life The Best Tips On How To Keep This Killer At Bay Discovering that you or a loved one has cancer can be utterly terrifying. All the same, once you comprehend the causes of cancer and learn how to reverse those causes, you or your loved one may have more than a fighting chance of beating out cancer.

Get My Free Ebook


Post a comment