Devita, Hellman, and Rosenberg's Cancer

LWBK1006-16 LWW-Govindan-Review December 12, 2011 18:55

192 DeVita, Hellman, and Rosenberg’s CANCER: Principles and Practice of Oncology Review

Answer 16.7. The answer is D.

There is no biologically inherent relationship between response and sur-

vival. For example, a meta-analysis on randomized clinical trials by Torri

et al. showed that a large improvement in response rate only corresponded

to a minor increase in survival. In addition, survival end points are com-

parative in nature and require taking the distribution of potential prog-

nostic factors into consideration.

Answer 16.8. The answer is B.

Trials of cytostatic agents often choose time to progression as the primary

end points. When testing cytotoxic agents, the use of Simon’s two-stage

designs is limited to situations where outcome assessment can be made

shortly after patients are treated. These designs usually are not practical

for end points requiring a long period of follow-up.

Answer 16.9. The answer is C.

Cytostatic agents, such as antiangiogenesis factors, growth modulators,

or cancer vaccines, usually selectively work on molecular targets to mod-

ulate tumor environment. They may prolong patient survival without

causing tumor shrinkage. Therefore, time to progression is often used as

the primary end point for such trials.

Answer 16.10. The answer is C.

Option A represents a design that used growth modulation index for the

assessment of efficacy, option B describes a randomized discontinuation

design, and option D is frequently used for trials on the combination of

cytotoxic agents. Simon’s Phase 2.5 designs are similar to Phase III trials

except that they allow a relatively large type I error and are able to identify

only a relatively large difference.

Answer 16.11. The answer is D.

Validated quality-of-life instruments are useful parts of Phase III trials,

especially if they are directly related to areas of concern to patients. Many

areas of patient benefit (e.g., longer survival, lower risk of relapse, acces-

sibility, or lower cost) may not be readily measured by these instruments.

Answer 16.12. The answer is B.

A random sample helps to provide a fair, unbiased comparison of the

treatments within the population of patients from whom participants

are recruited, although this population may not be representative of all

patients with the disease. Randomization may not result in equal, or

nearly equal, numbers of patients with a particular characteristic in each

treatment group, but it does distribute biasing characteristics according

to a known probability distribution so that they do not obscure identi-

fication of treatment effects, if any, by statistical tests. Matching is not

a substitute for randomization; in fact, matching too closely can mask

real treatment effects. Blinding of participants or treating physicians to