The problem of chemotherapy (or targeted therapy) is not its efficacy. It is the fact that you get a good response in some patients and poor response in others. During a clinical trial, if you do not select the patients according to a biomarker for inclusion, you will see a statistically significant effect only if the responding patients are not masked by the not responders. Last and not least, if the treatment is toxic you should desire to reserve the treatment only to people who will benefit from it.

I had in my careers during clinical trials experimentation very compelling evidences that a drug was working. We got responses in patients in fourth fifth line, never seen before in this clinical setting. However, when the pooled clinical analysis was conducted, the drug was killed since it did not reach the threshold of statistical significance. So the few responders were lucky (two of them are still alive to my knowledge).

Therefore, not only you need a biostat to properly design the clinical trials. You have to be sure that you are treating the right patients (and not treating the wrong ones). This is the only tool which can increase the probability of clinical success for new anticancer agents.

An academic mathematics / statistics group is basically always involved in clinical trials now.

From a biological point of view, the conventional chemotherapy destroys all dividing cells good or bad, but this does not mean it will sooner or later kill the patient by the side effects, as there are difference between good and bad cells, and difference among patients. A good point from cristino is that you have to treat the right patients (and not the wrong ones). Biomarkers are therefore needed for the right choice. A retrospective evaluation on those clinical trials, for instance, the one cristino mentioned might give very useful resources for the discovery of biomarkers.

Please correct me if I am wrong, but the whole point of clinical trials is to tell the responding patients from the non-responding ones. Only once this is established, then dependent on the result the patent owner can decide to apply for FDA approval for commercialization. Like Cristiano already indicated, when the number of responding patients is too low, the tried therapy is killed. Too bad for the few who would benefit. I am intrigued though how a toxic substance introduced to the circulatory system can benefit any patient. It makes no sense. We start clinical trials when some patients seem to benefit from a given treatment, but then we find it toxic to a beforehand unknown portion of the population. No surprise for a toxic substance! The problem is that in the past toxic substances like cisplatin were prescribed without any knowledge at all about its side effects, and without any knowledge at all about the etiology of the disease. The only thing that was known was that the cisplatin was toxic as it arrests mitosis. Good to treat (certain types of) cancer, but bad for the immune system, haematopoiesis, CNS stem cells, reproductive system and organ maintenance. Positive outcomes were measured with 5 year survival, so one may die right thereafter without putting a smut on the therapy.

Moving on, I would love to see how we currently look at the clinical trials conducted 10-20 years ago. Surely, today clinical trials are set up much more robust, but we still suffer from the aftermath of using toxic substances to "cure" patients. If only these substances were conjugated to molecules (read antibodies) to bring it to the cancer cells and nowhere else, then we could name such therapy targeted. But putting a toxic substance in the bloodstream is not a targeted therapy at all and it puts a lot of patients at fatal risk. Can anyone tell me how robust those original studies really were?

Maybe tomorrow we will discover new and not toxic treatments and we will be extremely happy of such outcome. However, nowadays we cannot get rid of the "chemo" we are using since the 60`s. There are very nice examples coming from history but I would like to give you some numbers and make an example in advanced ovarian cancer (stage III or IV).

A patient treated only with surgery (no chemotherapy) has a median survival time of 25 months. A patient treated with surgery plus chemotherapy has a median survival time of 62 months (at least here). These numbers are coming from large meta-analyses and the p value is extremely significant. This justify the use of chemotherapy in ovarian cancer. However, if you zoom inside the treated population you will discover that around 20% of the treated patients will relapse during or shortly after chemotherapy (< 3 months). These unlucky women will have a median survival rate of 24 months. In few words they will not benefit from chemotherapy. Why does a toxic agent work in cancer? The reason is simple. The drug kills more selectively the cancer cells than the normal cells. As I tried to explain the problem is not if chemo works or not. The above mentioned example is a clear proof that it is working. The problem is that it does not work for all the patients in the same way. This is also applicable for targeted agents, where unfortunately the efficacy is lower and the number of patients who will respond to the drug even lower. We have to switch the efforts from finding new treatments to the application of treatments to the right patients. I am convinced that we have many effective treatments which we are not using. In this context the clinical trial design becomes essential to proof the concept. But if you do not select the patient to treat with a strong rationale and a very robust assay, your efforts will be punished. In my humble opinion, this is one of the main cause of failure of very promising drugs killed in wrong trials in the last decades.

Dear Jan, "these substances were conjugated to molecules (read antibodies) to bring it to the cancer cells and nowhere else" is doing by many institutes. It is one of targeted therapies. However, results from my own experiments (including many ovarian cancers) and many others had shown those resistant cancer cells mentioned by Cristino could not be killed at all by the chemotherapy used in clinic in vitro, so even you bring those substance to the cancer cells via conjugated Ab that targeting specific cancer cell markers in vivo, those cancer cells are still unlikely to be killed. "conjugated Ab" approach that was reported many years ago is still ongoing, but not often with the conventional chemotherapy. Regards.

Dear Cristiano and Jimbiao, Many thanks for your valuable contributions so far. I would be the first one to call for substantial increased funding for fundamental research in order to identify new biomarkers enabling stratified medical care. I completely agree with Cristiano that clinical trials will have to be based on solid science. As a representative of the research antibody market, this is what I stand for! The boost in scientific research will not be financially possible until funds get shifted away from futile clinical trials (I would hope not all of them are futile) towards intensified molecular research. Any disease that is not properly understood will not yield adequate medication through trial and error. This is how chemotherapy came about and some cancer types (primarily ovarian cancer and the many types of lymphomas apparently) benefit to some extent (based on delay in death, rather than cure?). However, I see chemotherapy massively used all over the world for all sorts of cancer, not necessarily according to the latest science (non-academic, SOP-driven hospitals). For this reason I put in question the claims from trials done possibly decades ago based on which hospitals today still prescribe this type of treatments on a routine basis. The argument goes that we have nothing better at hand. This remains the case as long as funds for trials are based on targets with generic functions that happen to be involved in cancers as well and without proper insight in the etiology. Only science brings adequate solutions.

I would love to hear from other physicians and oncologists with strong views on this subject, preferably disagreeing with me (but some support is welcome too!). Can anyone answer my questions regarding historical trials on cisplatin and derivatives and possibly on taxanes as well?

Yes, ChemoTherapy would kill the patient, but only if you give it in unlimited doses and for an unlimited time period, but the same result would be obtained by drinking wine without a limit, as a matter of fact, Amy Winehouse died because an Ethyl poisoning; just the 'Therapeutic window/margin', ie, the difference from the active dose to the lethal dose is different.

Old CT, with exceptions as L-Asparaginaes, Anti-Estrogens, and so on, killed Cancer Cells by non-targeted mechanisms that affected also normal cells, and affected normal cells in a higher degree if it were in active replication, the subject of incorporating statistics in the design of Clinical Trials is not a simple one: of course, the ones knowing more about statistics are mathematicians, but it would be hard for them understanding the basis of disease and its biology, that are important in the designation of endpoints and other important issues in the planning of Clinical Trials, besides this, when engaging in research you always have the possibility of meeting the unexpected and the unknown, and this has also associated statistics problems.

Simple problems are solved by computers, complex ones require multi-disciplinary information, but having someone in command and making decissions is more efficacious than brainstorming and meetings, according to recently published evidence, and in dealing with disease, MDs should be in command. Salut †

Dear Jose, MDs decide on the best treatment for each individual patient, that is their job. But without solid science, such decisions are difficult to make. Here is a citation from >120 years ago:

"If it were not for the great variability among individuals, medicine might be a science, not an art"

Sir William Osler in The Principles and Practice of Medicine, 1892

When VAT was introduced in Spain, and until those requesting Medical consultations were exempted of paying it, some professionals requested being considered, for taxation purposes: 'Healer artists'. Unfortunately, few courts would admit for a suited MD, that the defense of his actions be: 'Medicine is an art, not a science, thus, you can't request results the way you can request results from the producer of spare parts for your car'.

Arts are beautiful, and it's beautful having your patients improved or healed, but I'd say a great step towards Medicine being mostly, if not only a Science, can be considered in the fact that the W Bargmann Histology text I bought in 1969 when entering the UNIZAR medical school, stated that: 'Lymphocytes have no known function'.

Salut †

When I was in Graduate School back in early seventies, our Chairman ,Jhon Pazur,used to say" if you have to use statistics to prove a point it is not worth proving"

I any trial efficacy should be viewed within the total circumstance and not evaluated by a preconceived standard. It is true that chemotherapy eventually kills patient's normal cells ,but in reality which is better: to kill some of patients cells or the patient. What I worry about now is use of any type of treatment against common sense wisdom and withholding any treatment when it makes sense.

I have been invited by Jan Voskuil to join the discussion on the efficacy of chemotherapy. My background is in immunology, radiobiology, bone marrow transplantation and radiation oncology, in the lab, as well as in clinical medicine.

Chemotherapy has enabled very significant progress in liquid tumors, some gonadal tumors and some pediatric tumors. This progress has come without having the full picture of which genes and which post translational mechanism control tumor cells and also without a complete picture of why chemotherapy can be curative in the tumors listed above and not in measurable solid tumors in adults, -measurable meaning detectable with the currently available diagnostic methods.

If I understand Jan's arguments right, he opines that solid cancers can only be cured if we have complete knowledge of the carcinogenic factors and normal tissue and cancer cells susceptibility to chemotherapy. If we have this knowledge we would be able to design curative chemotherapy, that does not kill the patient.

Jan and I agree that this is not the case at this time.

What can we do till the scientific community has reached the nirvana described above?

More money for fundamental research is important. Creative investigators are important.

Clinical Studies of the pharmaceutical industry have been expensive and essentially unproductive.

My preference is for doing the doable in brief and discriminating, 'mechanistic' phase 2 studies. Doing large multi-tarm, double blind studies are going to burn millions of dollars and as far as I know very few solid tumor patients are being treated today based on the results of a phase 3 trial. In our present state of ignorance we should do small studies and ask the right question. I do not agree with Jan that one cannot do simple studies, if we do not have 'the complete picture' of a given disease/cancer.

Think of the introduction of Salvarsan, peniciline, insuline, or the treatment of TB, beri-beri by investigators that did not have the complete picture.

Let's define and do the doable.

Many thanks for your contribution Huib! I fully agree, even to do simple experiments while not having the complete picture. My concern is that the clinical trials I have been seeing were far from simple and based on premature evidence. I am all for further studies to get better knowledge on how a candidate target is involved in a certain type of cancer. That is different from doing large scale trials on inhibitors to seek efficacy.

Anyway, my question still stands on whether historical clinical trials on chemotherapies based on which hospitals prescribe today meet today's more stringent standard. In addition, I wonder to what extend chemotherapies are used for solid tumor treatment, given they seem to have efficacy in liquid tumors only (based on Huib's comment)?