According to respected CancerGuide author Steve Dunn on his site's essiac information page:

So if there isn't good scientific data, what about anecdotal cases? I think that that cases can provide evidence in a limited way (See my Article on Evaluating Alternative Therapies for more on this.) Many of the articles I have read cite a few anecdotal cases, but in my experience most of these cases are either inadequately documented or, (and this is quite common) the patient had concurrent conventional therapy. Often, the conventional therapy is played down as something which could not possibly have been responsible for the good result, and in fact it is often true that such a good result would indeed be quite rare with conventional therapy. But even when the prognosis with conventional therapy is quite grim, a few, sometimes a very few, patients may have an extraordinary response. Given that Essiac is so popular, it isn't surprising at all that there are patients with a poor prognosis who used Essiac along with conventional therapy and who did much better than expected. Although it is also possible that Essiac did help these patients, you really can't tell either way, so this type of case is not good evidence for Essiac. While there is probably little harm in using Essiac tea, no one should rely on Essiac as a primary therapy for cancer -- the evidence just isn't there.

  Steve Dunn has a valuable criteria in his article for Evaluating Alternative Therapies.

Limitations of Cases and Testimonials There are two problems with testimonials. The first is the numerator/denominator problem. Even if you are convinced that some patients have been helped, you have no idea how many were not helped. Even if a treatment works rarely there may be an impressive number of testimonials or cases. This is a fundamental limitation of relying on testimonials. Nonetheless, testimonials can establish that a treatment works some of the time. The second reason that testimonials can be misleading is that there may be another explanation for the patient's survival. In fact, the vast majority of cases and testimonials I have looked into are not valid evidence - even when the patient is convinced that the therapy saved his or her life. In order to show effectiveness in a single case, you need to know that: The patient definitely had biopsy/pathologically confirmed cancer. You can't know if someone got better unless you actually know they were sick in the first place! As an example, I was told of a patient who was "cured" of a brain tumor by an alternative treatment that seemed scientifically absurd to me. I later found out that the patient has been "diagnosed" by a quack method called "applied kinesiology." There had never been an X-Ray or scan that showed tumor, much less a biopsy. Obviously, this patient never had a brain tumor in the first place. A biopsy is necessary to have a solid diagnosis of cancer. Most conventionally diagnosed patients do get a biopsy, but this is not always the case. Even patients who are diagnosed by doctors using conventional methods can be misdiagnosed. While this may be uncommon in general, it might be a common explanation for unexpected recoveries from cancer. Therefore, it is important to be sure that the diagnosis was biopsy confirmed. The patient definitely had observable tumor in their body when treatment started. Although spontaneous remission happens, it is quite rare in general, and exceptionally rare except for kidney cancer, melanoma, neuroblastoma, and lymphoma, so knowing that a patient had tumors that went away or shrunk under treatment is evidence that the treatment did something. If a patient had all disease removed by surgery or other treatment, then the fact that they stayed well afterwards may only indicate that the surgery or other treatment cured them. I have talked to people who had a locally advanced cancer removed who swore they were cured by an alternative therapy. Although their prognosis was indeed relatively poor, there was no reason to believe they were not just lucky. No conventional treatment could possibly have resulted in the patient's survival. Again, even if a patient was told chemo probably won't help, and then had a great response while also taking an alternative treatment, you can't be sure they were not lucky enough to have an exceptionally good response to the conventional treatment. There is objective evidence that the patient actually got better. This means either observed shrinkage or disappearance of tumors on x-rays, or on physical exam conducted by a physician, preferably independent of the alternative clinic. I would also accept extremely long survival (Many years) without symptoms in patients who had growing tumors before starting the alternative treatment and who would have been expected to die much sooner. A few alternative practitioners have developed their own questionable tests for cancer, and you should never accept the results of such tests as proof that there has actually been a response to treatment. The case is not just sales material! I have seen altogether too many vague testimonials on the Internet put forth by people who are interested in selling a product, often as part of a multi-level marketing scheme, but who know nothing about cancer or cancer treatment. These cases are almost always absurd or completely lacking in details. My advice is to ignore all such sales pitches. Since different cancers often respond to the same treatment very differently, it is important that you find evidence that the treatment works in your particular cancer, if at all possible (this may be difficult). Independent Qualified Review of Cases as Evidence You can see how tricky it is to determine if a case is really valid or not. It is far better if a qualified expert who is independent from the proponents of the therapy has already examined cases and found them to be valid. Sometimes, such "best case" evaluations do exist. Such an evaluation should include summaries of each case and you should be able to tell if the criteria I just described have been considered. You also need to consider is whether the expert is really independent. Such best case series still do suffer from the numerator/denominator problem.

--- Permission for use of material granted by Steve Dunn, CancerGuide, cancerguide.org.