This Article Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Cannistra, S. A. Search for Related Content PubMed PubMed Citation Articles by Cannistra, S. A. Social Bookmarking                            What's this?

When Is a "Prognostic Factor" Really Prognostic?

Beth Israel Deaconess Medical Center Boston, MA

I RECENTLY performed a MEDLINE literature search combining the terms "prognostic factor" and "cancer," which yielded no less than 2,393 results. The fact that this number seems out of proportion to the number of clinically useful prognostic factors in the field of cancer suggests that the majority of alleged prognostic factors have uncertain clinical and/or biologic value. Nevertheless, there is an undeniable importance in being able to determine which patients will do well without further therapy (the definition of a prognostic factor) and which will do well with some types of treatment and not others (the definition of a predictive factor). In the search for better biologic end points that could further refine our ability to prognosticate, the field has centered predictably on several types of molecular markers, including those that analyze proliferation (eg, Ki-67), metastatic potential (eg, matrix metalloproteinase expression), angiogenesis (eg, microvessel density), and, most recently, apoptosis. It is this last area of investigation that has been studied by our group in ovarian cancer, as well as several others, and that is the topic of an article by Baekelandt et al published in this issue of the Journal of Clinical Oncology.¹

Apoptosis is a genetically encoded program of cellular suicide, often triggered by a noxious stimulus such as cytotoxic chemotherapy. It is now widely accepted that most, if not all, forms of chemotherapy and radiation exert their lethal effects by causing enough cellular damage to activate the apoptotic program, which results in the destruction of vital cellular components by enzymes called caspases. At present, 13 caspases have been identified, each with its own substrate specificity and set of intracellular targets, and each representing an important downstream effector arm of the apoptotic program. Activation of the apoptotic program usually begins at the level of the mitochondrion, which releases cytochrome c into the cytoplasm in response to cellular damage. Cytochrome c, in conjunction with a protein known as apoptosis protease–activating factor 1, activates caspase 9. The subsequent cascade of downstream caspase activation is reminiscent of the clotting pathway and leads to the activation of several other caspases, which form the "business end" of the apoptotic process.² Thus, mitochondrial cytochrome c release is one of the earliest molecular signatures of most forms of apoptosis. The central role that the familiar BCL-2 proteins play in the apoptotic pathway stems from their ability to regulate the mitochondrial release of cytochrome c. In fact, the majority of BCL-2 family members are localized to the mitochondrial membrane, where some proteins, such as BCL-2 itself, actually protect the cell against the release of mitochondrial cytochrome c, whereas others, such as BAX, promote cytochrome c release. Thus, the balance between intracellular levels of antiapoptotic proteins such as BCL-2 and proapoptotic proteins such as BAX determines the threshold for mitochondrial cytochrome c release after an apoptotic stimulus.^3,4

There is now a significant amount of in vitro and preclinical data to suggest that manipulating the apoptotic threshold by altering the expression of BCL-2 proteins can affect the ability of chemotherapy to induce cell death. At the risk of oversimplification, upregulating BCL-2 or BCL-X_L generally makes cell lines more resistant to chemotherapy, and upregulating BAX often makes cells more chemosensitive.^5,6 The hypothesis that chemoresistance may be partly mediated by cells with an antiapoptotic phenotype (eg, cells that overexpress BCL-2 or BCL-X_L and/or that underexpress BAX) has gained significant conceptual appeal, and it has now been put to the test in a variety of tumor types, including ovarian cancer. In retrospect, the simplicity of this model almost predicts that it must be too good to be true. Nevertheless, several groups have produced data that nicely fit this paradigm, including a previous retrospective study in which we demonstrated an association among high levels of BAX expression, favorable response to paclitaxel-containing chemotherapy, and improved disease-free survival in patients with epithelial ovarian cancer.⁷ The fact that BAX expression retained independent prognostic significance in multivariate analysis strengthened the biologic implications of this association, but it does not remove the need to prospectively validate these results with larger numbers of patients. Prospective validation of a prognostic factor’s value in a different set of patients is an important way of confirming the clinical importance of any new tumor marker.

The recent study by Baekelandt et al¹ adds to the list of retrospective studies evaluating the prognostic value of BCL-2 proteins in ovarian cancer. These investigators measured expression of BAX, BCL-2, BCL-X_L, MCL-1, and p53 (a positive regulator of BAX transcription) in 185 consecutively treated patients with stage III epithelial ovarian cancer who received platinum-containing chemotherapy (non-paclitaxel based). BAX staining was scored in a graded fashion, with positivity being assigned if as little as 1% of cells expressed protein by immunohistochemistry. Clinical response was assessed using second-look laparotomy, although we are not told precisely how response was defined. The median follow-up was 85 months, certainly long enough to confidently assess the relationship between expression of a putative prognostic factor and overall survival. The incidence of BAX expression in the Baekelandt series is similar to our own (approximately 66%), although several important differences were observed. First, while it seems that BAX expression correlated with improved survival in the current series, this association did not retain independent significance in multivariate analysis. One wonders whether the fact that BAX-positive patients were more likely to have low-grade tumors with less residual disease is partly responsible for this phenomenon. More importantly, there was no association between BAX expression and chemoresponse, which raises the question of how, exactly, is BAX working to improve outcome in this series? The biologic reason for a lack of association between BAX expression and chemoresponse may relate to an earlier observation from our group, in which BAX was found to sensitize cells in response to some drugs, such as paclitaxel and vinca alkaloids, but not traditional alkylating agents, such as platinum.⁶ In fact, it has been recently shown that BAX may be partly mediating its effects by increasing the intracellular concentration of paclitaxel through a mechanism that is yet to be defined.⁸ Regardless of the mechanism, it is possible that the predictive power of a molecule such as BAX may become evident only in the context of paclitaxel-based chemotherapy, which was not used in the present study. Finally, in an attempt to derive a profile of markers that might confer independent prognostic significance, the authors identified a good prognostic group characterized by low-volume residual disease, absent p53 staining, and expression of both BCL-2 and BAX. This subset comprises only 20 patients and has a median survival of 104 months. In fact, the survival of these patients is suggestive of what I might have predicted for patients with stage IIIA disease, and one wonders how much overlap exists between this molecular profile and the 15 stage IIIA patients presented in Table 1 of the article.

Combining a series of factors such as residual disease status, p53, BCL-2, and BAX, and then reanalyzing the data in hopes of deriving an independent prognostic association, is a form of subset analysis that is fraught with danger, especially when only 20 patients form the subset. There may be other variables, some of which have been mentioned above, that are responsible for the good prognosis and that confound interpretation of the data. The fact that a combination of molecular markers may now have independent prognostic significance (compared with the use of single markers) is not as important to me as being able to separate patients into outcome groups in a manner that is superior to my clinical judgment alone. I am not convinced that the present study has accomplished this goal, although it represents a laudable attempt to do so. In ovarian cancer, it would be useful to identify the small subset of patients with early-stage disease who will relapse, so as to avoid unnecessary treatment of those who will not. For the more common scenario of advanced ovarian cancer, it is more important to identify the small subset of patients who will not relapse, since for these patients, standard therapy with paclitaxel and carboplatin is sufficient. For everyone else, treatment is indicated, preferably in a clinical trial setting, and all of the prognostic factors in the world will not alter what we do.

Perhaps the most important contribution of a study like that of Baekelandt et al is not to develop another set of prognostic or predictive factors, but to provide clinical insights in support of a biologic hypothesis. Any data that support the biologic relevance of markers such as BCL-2 or BAX might lead to novel ways with which we might overcome chemoresistance. In this regard, we have recently shown that tumor-selective expression of the BAX gene provides a powerful death stimulus for ovarian cancer cells in preclinical models,⁹ an observation that supports some of the data provided in this study as well as others. However, given the complexity of molecules involved in the apoptotic pathway, it is doubtful that investigating a small subset of BCL-2 proteins will yield the kind of information that we really need to refine our prognostic abilities as clinicians. The approach used in the present study is dependent upon our ability to determine which molecules are important to study, and this in turn is limited by the "glass ceiling" imposed by our relative ignorance of the basic biology of cancer. Therein lies the power of microarray technology, which will be the subject of an upcoming Biology of Neoplasia review, since this technique has the ability to define molecular profiles without biologic prejudice, without the knowledge of how genes may interact to actually confer a worse prognosis. In the future, it is likely that our quest for prognostic precision will come full circle and that we will be able to foresee who will relapse, and who will not, without even understanding the biologic significance of the genes expressed by the tumor. Until then, good clinical judgment, seasoned now and then by the occasional molecular marker (eg, estrogen receptor expression in breast cancer), is still the mainstay of determining prognosis in patients with epithelial cancers.

REFERENCES

1. Baekelandt M, Holm R, Nesland JM, et al: Expression of apoptosis-related proteins is an independent determinant of patient prognosis in advanced ovarian cancer. J Clin Oncol 18: 3775-3781, 2000[Abstract/Free Full Text]

2. Reed JC: Dysregulation of apoptosis in cancer. J Clin Oncol 17: 2941-2953, 1999[Abstract/Free Full Text]

3. Yang E, Korsmeyer SJ: Molecular thanatopsis: A discourse on the BCL2 family and cell death. Blood 88: 386-401, 1996[Free Full Text]

4. Chao DT, Korsmeyer SJ: BCL-2 family: Regulators of cell death. Ann Rev Immunol 16: 395-419, 1998[Medline]

5. Reed JC: Bcl-2 family proteins: Regulators of apoptosis and chemoresistance in hematologic malignancies. Semin Hematol 34: 9-19, 1997[Medline]

6. Strobel T, Swanson L, Korsmeyer S, et al: BAX enhances paclitaxel-induced apoptosis through a p53-independent pathway. Proc Natl Acad Sci U S A 93: 14094-14099, 1996[Abstract/Free Full Text]

7. Tai YT, Lee S, Niloff E, et al: BAX protein expression and clinical outcome in epithelial ovarian cancer. J Clin Oncol 16: 2583-2590, 1998[Abstract]

8. Strobel T, Kraeft SK, Chen LB, et al: BAX expression is associated with enhanced intracellular accumulation of paclitaxel: A novel role for BAX during chemotherapy-induced cell death. Cancer Res 58: 4776-4781, 1998[Abstract/Free Full Text]

9. Tai YT, Strobel T, Kufe D, et al: In vivo cytotoxicity of ovarian cancer cells through tumor-selective expression of the BAX gene. Cancer Res 59: 2121-2126, 1999[Abstract/Free Full Text]

CiteULike    Complore    Connotea    Del.icio.us    Digg    Facebook    Reddit    Technorati    Twitter    What's this?

This Article Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Cannistra, S. A. Search for Related Content PubMed PubMed Citation Articles by Cannistra, S. A. Social Bookmarking                            What's this?