Individualizing cancer therapy

In this issue of Leukemia and Lymphoma, Ngo and colleagues describe the predictive value of cyclin D1 expression as a predictor of bortezomib response in multiple myeloma. The ability to predict treatment response based on genetic expression would be of value, if it could be confirmed in a larger study. The present publication should be considered hypothesis-generating, but the concept of having a diagnostic test that can predict a response to bortezomib as opposed to an empiric trial of therapy is potentially useful. The response rate reported in this paper is unusually high, and there is lack of a suitable control group [1]. Whether cyclin D1 predicts responsiveness to bortezomib specifically, or is a general indicator of good prognosis, would need to be verified in a group of patients treated with melphalan-based or immunomodulatory drug (IMiD)-based therapy.

Cyclin D1 is a protein that, in humans, is encoded by the CCND1 gene. The protein encoded by this gene belongs to the highly conserved cyclin family, whose members are characterized by protein abundance throughout the cell cycle. Cyclins function as regulators via cyclin-dependent kinases. Different cyclins exhibit distinct expression, which contributes to the temporal coordination of a mitotic event. Cyclin D1 forms a complex with and functions as a regulatory subunit of cyclin-dependent kinase-4 or cyclin-dependent kinase-6, whose activity is required for cell cycle transition. Cyclin D1 interacts with tumor suppressor protein RB, and the expression of CCND1 is regulated positively by RB found on chromosome 17. Mutations, amplification, and overexpression of cyclin D1, which alters cell cycle progression, is observed frequently in a variety of tumors, and has been felt to contribute to tumorigenesis. Overexpression of cyclin D1 is not limited to myeloma, but has also been implicated in the pathogenesis of a variety of hematologic neoplasias such as mantle cell lymphoma, hairy cell leukemia, and B cell lymphoma, as well as in other solid benign and malignant tumors including parathyroid adenomas and breast, lung, and colon carcinomas.

The expression of cyclin D1 is subject to both transcriptional and post-transcriptional regulation [2], and different pathways assume a dominant role in regulating its expression. This regulation can contribute to overexpression of cyclin D1 in the genesis of tumors, while nuclear factor κB (NFκB) pathways regulate cyclin D1 gene transcription.

Other groups have described the favorable impact of cyclin D1 overexpression on bortezomib responsiveness. In breast cancer, cyclin D1 repression of STAT-3 expression may explain the association between cyclin D1 overexpression and improved outcome in breast cancer. Bortezomib can amplify the pro-apoptotic function of cyclin D1, and may explain why increased expression can predict response to the drug. Conversely, tumors that fail to express cyclin A have been implicated as independent markers of tumor progression and poor response to chemotherapy in advanced head and neck cancers [3].

The concept of using gene expression to individually tailor treatment specifically for cancer patients or to predict the toxicity of therapies has been a long sought after but thus far elusive goal. Gene expression profiling has demonstrated that the ability to predict a response to radiation treatment can occur by measuring the labeling index after intravenous injection of iodinated uridine, or by estimating cyclin D1 expression [4].

In adult human T-lymphotropic virus type I (HTLV1) associated T-cell acute lymphoblastic leukemias, Alizadeh et al. have identified specific gene expression signatures associated with response to zidovudine (AZT) and interferon-α (IFN-α) therapy, the mechanism of which was attributed to repression of cell-cycle regulated genes [5].

In breast cancer, chemotherapy outcomes vary widely. Tumor proteomic profiles have been investigated for their ability to predict an individual tumor response to treatment. Breast cancer tumors were analyzed using surface-enhanced laser desorption ionization mass spectrometry (SELDI) (a classification model correctly identified most tumor responses with an accuracy of 92.3%). This suggests that breast cancer protein biomarkers can be used to preselect patients for optimal chemotherapeutic treatment [6].

The concept of tailoring therapy is probably best based on multidisciplinary modalities. In this respect, there is a role for imaging with positron emission tomography with 18F-fluorodeoxyglucose (PET-FDG) for patients with Hodgkin lymphoma disease and other lymphomas, as a decision tool used to customize duration of therapy [7].

The development of the field of epigenetics, such as using histone deacetylase inhibitor-based treatments, can be guided using gene expression profiling studies on lung cancer cell lines. A nine-gene classifier has been identified to be useful in predicting drug sensitivity to histone deacetylase inhibitors, and is felt to contribute to improved individualized therapy for lung cancer patients [8]. Microarray analysis can be used for the prediction of response to radiotherapy in human cancers. Some of the identified genes are associated with DNA repair, apoptosis, growth factors, signal induction, cell cycle, and cell adhesion. Global gene expression profiling of responders and non-responders can be used to predict response to radiotherapy [9]. The use of genomic predictors of response to cisplatin and pemetrexed can be incorporated into strategies to optimize therapy for advanced solid tumors, and provide a rational approach to the treatment of cisplatin-resistant patients with a positive predictive value of 78% and a negative predictive value of 100% [10]. A multiplex approach combining the different biological levels, DNA, RNA, and protein, may be necessary to functionally classify malignant tumors, resulting in optimization of targeted therapies [11]. Attempts to predict chemotherapy responsiveness have been demonstrated in Waldenstrom macroglobulinemia treated with rituximab, and FcγRIIIA (CD16) receptor expression is associated with a higher response rate to rituximab [12].

The use of pharmacogenomic techniques to predict outcomes is not limited to cancer. The long-standing practice of measuring the genetic polymorphisms for glutathione S-transferase and thiopurine S-methyltransferase in predicting the inter-individual differences in therapeutic response and toxicity to thiopurines in inflammatory bowel disease is well under way [13]. Methylenetetrahydrofolate reductase polymorphisms have an important impact on folate metabolism and the cytotoxicity of 5-fluorouracil. Two common, single nucleotide polymorphisms reduce the activity of the enzyme, and can be associated with lower toxicity in 5-fluorouracil treatment [14].

The pharmacogenetic potential of identifying polymorphisms in cytochrome P450 (CYP450) has been reported to impact outcomes in the use of psychiatric medications. Dosing recommendations based on variant alleles are appearing in the literature for various classes of psychiatric drugs, but no firm guidelines exist. The hope would be to identify individuals at the metabolic extremes with the highest risk for adverse outcomes [15].

The field of individualized treatment of our cancer patients is being researched [16], but implementation of routine pharmacogenomics and single nucleotide polymorphism analysis into individualized treatment remains premature, and will require further confirmation in large studies. The ability to ultimately offer specific chemotherapeutic agents in the right dose for a unique indication based on genetic profile offers promise for an improved therapeutic index for our patients.