Chemotherapy Sensitivity/Resistance Assays in Gynecologic Cancer: Are We There Yet?
By Robert L. Coleman, MD
State of Affairs
Over the last few decades, technology has existed to evaluate the cytotoxic effects of single-agent and combination chemotherapy on cancer cell lines derived from an individual patient's tumor specimens.1-4 However, investigators and clinicians have been frustrated as the fruit of this technology—a reliable and reproducible assay to help them treat their patients with the agent or agents most likely to benefit them—has yet to be proven. Currently, the determination of chemotherapy to be used individually is a decision made empirically; supported for the most part by clinical data generated ideally from randomized and non-randomized clinical trials on like cohorts of patients. However, since in all such trials, only a proportion of patients respond, the science is obviously imperfect and the decision subject patients to potentially toxic therapy that, on a individual basis, may have little chance for success. Tailored chemotherapy remains an important and extensively sought after endpoint in cancer treatment, and as such, drives many clinicians to utilize some of the many currently available technologies that claim to provide some improved guidance over our best guess.
A chemotherapy sensitivity and resistance assay is a laboratory algorithm wherein a sample of human tumor is subjected, under experimental conditions, to various chemotherapeutic agents and concentrations in order to assess response (tumor survival). Two broad categories of assay-intent separate the available technologies: those that evaluate the inhibition of cell growth and those that address chemotherapy-associated cell death. While these intents appear similar, they are very different in their laboratory aim and may produce vastly disparate results.5 In most cases, several drugs and combinations are evaluated. Theoretically, the most active agent or combination could be picked (sensitivity assay) or eliminated (resistance assay) from an empiric program, offering a more precise decision tool. The hypothesis is that this maneuvering will benefit patients in the ultimate outcome, survival. While the concept is simplistic and rational, the effects of chemotherapy response and patient survival are complex and sometime counterintuitive. For instance, it is probably over-reaching to assume a limited sample of tissue obtained from either the primary or a metastatic site, at primary diagnosis or in recurrence and following prior chemotherapy or radiation exposure will be representative of active disease at any one time. Similarly, the relationship between response and overall survival is at best tenuous and reflects issues not measured in the lab such as toxicity, quality of life, and performance status.
Comprehensive discussion of the individual available assays is beyond the scope of this commentary but they will be categorically introduced for orientation. Most available assays evaluate isolated tumor cells from a tissue biopsy or fluid specimen after which the cells are incubated in the presence of a chemotherapeutic agent. Inhibited growth and/or cell death are end points allowing a sensitivity characterization. Other assays reach this determination by evaluating the ability of a chemotherapeutic or combination to kill a certain proportion of cells relative to baseline. Those agents reaching a specified cut-off are allocated as sensitive.
In the September 1, 2004 issue of the Journal of Clinical Oncology, the American Society of Clinical Oncology (ASCO) commissioned a Working Group6,7 "to develop a technology assessment of chemotherapy sensitivity and resistance assays in order to define the role of these tests in routine oncology practice." Collaborating with the Blue Cross and Blue Shield Association Technology Evaluation Center, the 2 groups independently evaluated the world's written literature, seeking to identify clinical trials which assessed the ability of an assay to positively affect clinical outcome when it was incorporated in a decision process. Despite the plethora of abstracts dealing with the topic (over 1100), a surprisingly limited number of clinical trials (12) met a priori requirements of: prospective design, comparative outcome between assay-directed and empirically treated groups, sufficient sample size, and contemporaneously treated cohorts. While some of the trials did show various end point advantages, such as response or progression-free survival, superior overall survival was only inconsistently documented. The conclusion reached by both organizations was that the technology was not ready for prime time. However, conceding the importance of the concept in general, they called for prioritized investigation with inclusion of the technology in future prospective clinical trials.
Assay-Directed Therapy and Malignancies
While the conclusions reached from the ASCO Working Group appear sound, it is important to consider these conclusions in the context from which they were derived. The strict criteria for study inclusion limits available data to published series, some incorporating older and/or impractical technology. In addition, several studies, evaluated by the Group, demonstrated response and survival outcomes that were favorable or significantly improved for the assay-directed treatment cohorts. Many of these were among ovarian cancer patients. The applicability of assay-directed therapy may be better suited for gynecological malignancies, particularly ovarian cancer, given not only the large emporium of active agents, but also the significant duration of time one may have to treat a patient. However, without a well-designed, randomized clinical trial, generalized utilization of (and reimbursement for) assay-directed treatment, even among these patients, will not be realized.
For assay-directed therapy to make a real impact in the treatment of a disease, several obstacles, conditions, and challenges need to be met and overcome. First, the assay must return reliable results in a timely manner. A corollary is that in the majority of cases, an interpretable result is returned. In the case of ovarian cancer, many patients are ready for chemotherapy shortly after cytoreductive surgery or tissue biopsy. Although the relationship of early treatment after surgery to survival has been challenged,8 delay of therapy in a symptomatic, anxious patient may be difficult. Assay results that take weeks to return are likely to be of limited benefit if a delay in treatment initiation is required; even more so if the probability of a meaningful assay is low (less than 75%). In addition, the allocation of resistance and sensitivity needs to be meaningful, reproducible, and reflective of in vivo observations. Second, a 1-time biopsy is likely not to be accurately reflective of all disease conditions in which the assay may be intended. In reference again to the ovarian cancer model, tissue is usually available from primary debulking surgery, and an assay may be run on tissue from the primary, and a metastatic site to determine a rational drug choice at primary therapy. However, when the tumor recurs, does it retain the same phenotype? In addition, does acquired resistance from previous therapy alter subsequent assay allocations with other agents? For instance, would a recurrent ovarian cancer patient, not previously resistant to doxorubicin, but failing second-line paclitaxel, now become so in the acquiring of this drug-resistant phenotype? Likely, additional information such as molecular profiling will be needed to make better predictive inferences. Third, clinical trials evaluating the technology will need to take into account the likelihood that the assay-directed therapy will pick the empiric choice, or a regimen with fractionally lower response probability. In disease sites where the proportion of difference is less than 25%, a significant increase in patient numbers will need to be enrolled, unless the assay-directed survival difference is large (not likely). Likewise, if drug A is the most sensitive, and produces a response rate of 50%, but drug B (empiric) produces a response rate of 45%, large accrual cohort will be needed, and a decision as to whether this difference is clinically meaningful will need to be addressed. Since in many gynecologic cancers the number of active agents may be considerable, the number of combinatorial options produces a highly complex and factorial set of possibilities; some up front prioritization will need to be determined for trial design. Lastly, in the event that no agent is active in a particular patient, clear validation of the assay's predictive nature needs to be explored. Potential for harm may be present in the event a patient is denied standard therapy based on an invalidated negative assay result.
The appeal of developing a tailored therapeutic program that offers a patient the best chance to live the longest or cure their disease is great, and will always drive the search for inventing a better cog in the decision process over empirical therapy. It is time to incorporate these assays into a well-designed clinical trial, preferably randomized with attention paid the challenges outlined above. Simplification, sophistication, and availability at a reasonable cost will be necessary, as well for the assay-directed approach to enter mainstream cancer care. Development in genomics and proteomics, as well as microarray technologies will likely also help solve this clinical puzzle.9
1. Loizzi V, et al. Am J Obstet Gynecol. 2003;189: 1301-1307.
2. Gazdar AF, et al. J Natl Cancer Inst. 1990;82:117-124.
3. Stratton JA, et al. Gynecol Oncol. 1984;19:336-347.
4. Greenberg PL, et al. Cancer Res. 1977;37:1794-1799.
5. Heppner GH, et al. Cancer Res. 1978;38:3758-3763.
6. Schrag D, et al. J Clin Oncol. 2004;22:3631-3638.
7. Samson DJ, et al. J Clin Oncol. 2004;22:3618-3630.
8. Sorbe B. Int J Gynecol Cancer. 2004;14:788-793.
9. Tanaka T, et al. Int J Cancer. 2004;111:617-626.