Critique of Annonymous Reviewer ("The Article Reviewed")
Commentary on Current Status of Cell Culture Resistance Testing
Weisenthal and Nygren provide an overview of various methods of drug
resistance testing, and of publications supporting their role in cancer
treatment. With nearly a decade and a half since the original attempts
at predictive assays pioneered by Salmon and Hamburger, and the early publications
by Weisenthal, much has transpired in terms of methodology and clinical
applications, although rigorous clinical trials have rarely been conducted
to validate the concept.
To support this last statement, I shall expand on the possible role of predictive assays in the treatment of ovarian cancer-one of the preferred targets discussed by the authors. My purpose is to bring out certain clinical issues that emphasize reasons for considering the data presented as preliminary and not ready for wide applications in patient care. Specifically, the following aspects of the data should be viewed from the perspective of state-of-the-art treatment of ovarian cancer:
1. Correlation of assay findings with response: outcome in ovarian cancer is well known to be dependent on a number of pretreatment characteristics, that vary somewhat in relation to whether the end-points are response rates, time-to-failure or survival. For second-line therapies these have consisted of bulk of disease and performance status among several variables examined (Eisenhauer, Safra). For both first-and second-line sensitivity to a platinum are likely to have impact on outcome. The series listed in Table 1 under ovarian cancer do not provide this information. Three of the series (Blackman,Ohie,Sevin2) have studied samples from untreated patients, and the true positive (TP) rate is close to 50%, with a low (12%) false positive (FP) rate. The other series include patients failing prior treatment, and the TP rate is 30% with a somewhat higher FP rate. The true negative (TN) rate increases sharply (from below 20 to near 50%) when dealing with less responsive disease. All series fortunately suggest a low rate of false negatives. Overall, the data are consistent with clinical findings in relation to platinum sensitivity prior and after treatment. They also hint to great problems in obtaining a reliable answer from an assay in the more refractory setting -- not an unexpected finding. (click here for Weisenthal's rebuttal to this paragraph).
2. Clinical trials: only one clinical study is quoted, and it showed a strikingly better outcome in 25 patients with assay-directed, compared to 30 concurrent well matched controls (Kurbacher). The encouraging results stimulated the authors to begin a prospective clinical trial. At present, one needs to be more critical of this report: 1)the study is quite small, 2)any such comparison is still subject to bias, and 3) the impressive nearly 2 year (90 weeks) median survival overshadows the 50 week progression-free survival suggesting that these patients shared features predicting for good survival that did not necessarily reflect the treatment received. Pertinent to this last point, platinum sensitive patients have a longer survival (and also higher response rates) after treatment with drugs other than platinums. The same authors, however, describe in an abstract (ASCO '99 1384 abstr) additional patients treated with assay-directed combinations and indicate that platinum sensitivity in this series did not influence response, progression-free survival, or survival, and call attention to their confirmatory phase III study that is ongoing.
3. Response as a measure of benefit: the premise that the assay predicts response and that this relates to patient benefit is, of course, open to question. Again, in first line treatments based on platinums, their superior response rates are associated with substantial gains in progression-free and overall survival in a number of randomized trials. However, the situation in 2nd line or previously treated patients is far from clear. Even assuming that the assays are able to accurately predict for responsiveness to certain drugs, it is not clear whether such responses will have an impact in survival and be superior to treatment with other drugs agents that may not produce equivalent objective response rates. Only a few 2nd line therapy trials have been performed that suggest this treatment to yield a superior survival even in the instance of platinum drugs that yield substantial response rates in pretreated patients who are platinum-sensitive.
4. Cost/benefit analysis: a clinician concurring with obtaining an assay directed assessment is faced with a number of questions difficult to answer. a)what is the likelihood that that the answer will prove helpful?, b)what is the cost of the procedure?, and c)is it worth the added morbidity required in some instances to obtain the sample? These questions are not unlike those raised by the authors of the overview. Their claim that too high a bar is placed against proponents of assay-directed decisions must be contested. Aren't the usual requirements of clinical trials that they address risk versus benefit questions, avoid reliance on anecdotal reports, and generate the appropriate studies to prove hypotheses? In this context, the case for assay-directed determinations to select the treatment of patients with ovarian case has not yet been made. (click here for Weisenthal's rebuttal to this paragraph).
Concluding remarks: In spite of the evidence reviewed in favor of implementing assay-directed selection for the treatment of ovarian cancer, more information is required before this can be accepted as helpful in routine management. The results presented in terms of response (True positives) do suggest some validity in detecting drug activity, but let us examine how this might lead to patient benefit in ovarian cancer. In first-line treatment, it would be risky to abandon cisplatin or carboplatin for another set of drugs that might yield responses, but do not have the curative potential nor the same probability of durable responses associated with the use of platinums. Current results of platinum-based chemotherapy (empirically determined) yield response rates in excess of 70% and a clinical trial to exceed these results would be most difficult to design. For Retreatment, the selection of therapy is guided by the likelihood that responsiveness to platinums persists. A 6-month relapse-free interval has been arbitrarily used to distinguish sensitive from resistant although a more sophisticated six category system has recently been introduced. If one accepts that the patient is potentially platinum-sensitive, the assay might indeed prove useful in confirming or disapproving this potential. In this latter case , platinums could be avoided and toxicity would be spared in circumstances that cures are highly unlikely. A clinical trial is required to prove this premise and has been under discussion in the Gynecologic Oncology Group. On the other hand, if the patient is empirically deemed platinum-insensitive, it would be quite unlikely that the assay could be relied upon to greatly enhance the likelihood of response. Under these circumstances one might anticipate a low percentage (10?) of responses, and a substantial percentage of false positives from the assay.
The likelihood of combinations being more helpful when guided by assay results, this also seems unlikely since a vast number of studies even in front line have failed to consistently prove superiority of a combination over platinums by themselves.
Finally, the view that assay studies have yielded some striking therapeutic breakthroughs must be tempered by the anecdotal nature of the information. Gemcitabine+Cisplatin combinations have emerged not only through the assay, but also through empiric studies in ovarian and other cancers: and through mechanistic studies. Undoubtedly, with identification of new therapeutic targets, characterization of tumors for the presence or absence of such target will become commonplace. Within this broad context the author's plea for more rational selection is justified. However, I can think of only one scenario in ovarian cancer management where the assay might be reasonably included within consideration of other commonly used clinical procedures. This would be the instance of a patient who has her first clinical relapse leading to debulking after a free interval exceeding 6 months. But the lack of data on our standard practices such as secondary debulking is a weak justification for adding another practice without pursuing the appropriate trials.
Click here for Weisenthal's general rebuttal
to this entire review
Click here to return to Human Tumor Assay Journal Home Page