This story has been updated from a version originally posted on June 16.
Tamoxifen is among the most successful breast cancer therapies, but in about 25 to 35 percent of patients — including half of those who have recurrent disease — the selective estrogen receptor modulator marketed by AstraZeneca has no therapeutic effect.
In an effort to investigate why tamoxifen doesn't work on some patients, and, in the process, develop a method to predict who may not benefit from the drug, a group of researchers from the Netherlands and the US have performed a proteomics study to identify a putative protein profile associated with tamoxifen resistance.
Their work, built on an earlier study the researchers conducted using nano-LC-FTICR mass spectrometry combined with the accurate mass and time tag approach, identified 100 putatively differentially abundant proteins between tamoxifen-sensitive and tamoxifen-resistant tumors.
They also identified one biomarker, extracellular matrix metalloproteinase inducer, as the most discriminating protein of the putative profile. Levels of EMMPRIN were higher in tamoxifen-resistant tumors and the protein was associated with earlier tumor progression following first-line tamoxifen treatment as well as poor clinical outcome.
The work is described in a study published in the June issue of Molecular & Cellular Proteomics. In the paper, the authors said that because tamoxifen resistance is a "major cause" of death in recurrent breast cancer cases and current clinical factors are predictive of therapy response in about only half of all treated cases, the identification of proteins associated with tamoxifen resistance could be a "first step toward better response prediction and tailored treatment of patients."
There is extensive genetic research into tamoxifen resistance, and a number of studies have linked variations of the CYP2D6 gene to response to the drug. In fact, the US Food and Drug Administration has been debating since 2006 whether to change the labeling of tamoxifen so that users would get routinely genotyped before doctors prescribe the treatment.
In the MCP article, the authors acknowledge that proteomic research lags behind genetic studies in tamoxifen resistance, but note that "protein-level information is crucial for the functional understanding and the ultimate translation of molecular knowledge into clinical practice, and proteomics technologies continue to progress at a rapid pace."
Arzu Umar, the lead author on the article and a post-doc at Erasmus Medical Center, told ProteoMonitor that because cellular process are regulated on the epigenetic, transcriptional, translational, and post-translational levels, as well as by microRNAs,"this automatically means that changes found at the gene-expression level do not necessarily correlate with protein levels. Even more, it had been shown that gene and protein expression correlate rather poorly."
It is therefore important to know protein levels, especially because proteins are the functional building blocks of the cells, she said. "From a clinician's point of view, most diagnostic tests are based on protein assays, such as ELISA. If we find good protein markers in the lab, this can be more easily converted into a clinical test than gene-based information," she said.
However, she cautioned that while the proteins identified by the research may be associated with tamoxifen resistance, they may not necessarily cause it.
"This profile says something about the predictive value but it doesn't mean that all of them have to have a causative role," Umar said. "That is something that would need to be investigated — whether it is just a marker for it or whether it's really involved in it."
Umar acknowledged that other breast-cancer research has employed proteomics technologies such as 2D gel analysis or LC-MS for protein separation. However, she and her co-authors note that the proteomic make-up for a cultured cell is known to be "rather different" from that of a tumor cell "surrounded by its microenvironment." Cell lines lack the "required follow-up information for answering important clinical questions," and tumor tissues, particularly breast cancer tissues, "are very heterogeneous in the sense that they harbor many different cell types, such as stroma, normal epithelium, and tumor cells," they wrote.
In their approach, they used laser-capture microdissection technology, which they said has emerged as an "ideal tool" for the selective extraction of cells from their natural environment. LCM-derived breast cancer tumor cells have been used by other researchers, they added, and resulted in the identification of proteins involved in breast cancer prognosis and metastasis.
Those studies demonstrate that proteomics technology has advanced enough that "it can contribute to biomarker discovery," but the technology remains hindered by drawbacks such as the need for large sample requirements and low proteome coverage when small amounts of starting material are used.
"When you're working with very small amounts of sample like microdissected material, if you don't have a lot of starting material, and if you did a conventional mass spectrometry-based approach, you would probably not see a lot of proteins coming off the column in your mass spectrometer," Umar said. "So you want to use a method that is the most sensitive … that gives the highest resolution, the highest dynamic range."
In earlier work, she and her collaborators demonstrated that nano-LC-FTICR mass spec coupled with the AMT tag approach for proteomics characterization of 3,000 LCM-derived breast cancer cells improved proteome coverage over conventional techniques.
Using the same strategy on eight pools of tumor cells in duplicate or triplicate derived from 51 fresh frozen primary invasive breast carcinomas — 24 sensitive to tamoxifen therapy, 27 resistant to therapy — the scientists "put forward a putative protein profile that may predict the outcome for tamoxifen therapy in breast cancer patients," they said.
[ pagebreak ]
The results are preliminary, however, and "whether this profile as a whole is a good predictor for tamoxifen therapy response in a larger, independent group of patients, and whether it is applicable to chemotherapy as well will be the subject of further investigations."
Profile of Resistance?
In the meantime, more than half of the 100 proteins identified by their initial research had not been previously associated with tamoxifen resistance, with breast cancer, or with resistance to other therapies, Umar said.
Initially, she and her team were able to identify 17,263 peptides corresponding to 2,556 proteins through AMT tag database matching. Between the two sample sets of tamoxifen-sensitive and -resistant breast cancer tissues, 1,713 proteins, identified by 13,729 peptides, were identical.
The 1,713 proteins were subjected to statistical analysis in order to discover proteins associated with tamoxifen resistance: The univariate two-sample t test from BRB-Array Tools was used to search for differentially abundant proteins between the two sample sets, resulting in a list of 153 discriminating proteins using a significance threshold of p greater than .05.
The number of proteins was further winnowed to 100 when the proteins were subjected to a Wilcoxon rank sum test. These 100 proteins "were designated as a putative protein profile associated with the type of response to tamoxifen therapy," the researchers wrote. In this profile, 46 proteins had higher relative abundance in the sample set of 27 tumors that were tamoxifen resistant, while 54 proteins had higher abundance in the sample set with sensitivity to tamoxifen.
The top discriminating protein was splice isoform 2 of basigin precursor, also described in the literature as CD147 or EMMPRIN, which has been associated with other cancers.
To test the predictive power of the 100-protein putative profile, the researchers performed supervised hierarchical clustering and found that on their average relative abundances, the tamoxifin-sensitive and -resistant proteins were "effectively separated from each other."
In addition, they performed targeted LC-MS/MS analyses to verify the presence and abundance level of all profile proteins in separate tumor samples. Fifty-five proteins out of the 100 were verified either by MS/MS sequencing or by survey mass spectra, and retrieved quantitative data on 47 of the proteins.
EMMPRIN was not identified by this approach, however, and the authors attributed the "relatively low verification rate" to the use of different samples and different LC-MS platforms for the discovery and verification parts of the study. In the discovery work, the researchers used an FTICR platform on microdissected tissues, while for the verification phase, they used an Orbitrap on whole tissue lysates.
Umar and her colleagues, however, validated EMMPRIN by using an antibody directed against the C-terminal of the protein, which recognized all splice isoforms of EMMPRIN.
Using immunohistochemistry to validate differential abundance of EMMPRIN between tamoxifen-resistant and -sensitive patients, the researchers observed that none of the complete-remission tumors displayed any EMMPRIN staining, "whereas highest EMMPRIN staining was observed in [tamoxifen-resistant] tumors," they wrote.
Umar said that EMMPRIN may be an especially important protein because in addition to being implicated in other cancers, other studies suggest it may play a role in chemotherapy resistance, though there is "no clear description so far that it was involved in tamoxifen resistance.
"My idea is that it may be involved in therapy resistance, in general, whether it's chemotherapy or tamoxifen," Umar said, adding that that is an area of continuing research by her and her co-researchers.
They also are now in the beginning stages of developing MRM assays to validate their findings. They plan to start validation work on the 47 proteins for which quantitative data was available, although it may not be possible to validate all 100 proteins in the putative profile because an MRM approach may not be appropriate for some proteins and antibodies do not exist for all the proteins.
"You need to select good peptides and they would need to have good mass spectra, so having a good MRM assay depends on whether you can get mass spectra from those peptides and whether you can get good transitions," Umar said.
The researchers are building a tissue-specific AMT database, as well. "We believe that if we are matching with a tissue-specific AMT database, we will find many more tissue-specific proteins," Umar said, adding they are using breast cancer tissues to build that database.
Finally, she and her colleagues are trying to expand their discovery phase to a larger set of samples in which the samples would remain individualized so that pooled material wouldn't be used.