> Surveying the oncology landscape, what you see is that US society essentially feels that some cancers are simply more worth treating than others

Breast cancer has been favored in terms of press coverage and funding for quite some time now. It seems only with the advent of some successful, and profitable, targeted therapies that there’s been more attention paid, and more investment in research regarding, lung cancer.

Riddle me this, Batman: how can Progression Free Survival (PFS) be significantly increased in a population without a concomitant increase in Overall Survival (OS)? It just doesn’t seem to make sense. One would think that less progression begets longer survival.

Joe,

As you might imagine, Dr. West (aka Batman) has discussed this issue before: http://cancergrace.org/forums/index.php?topic=1030.0

Jim

Thanks for the link, Jim. Essentially, the idea is that the opportunity for subsequent effective treatments can allow equalization of the groups over time. You might well imagine that if the overall treatment is more of a marathon than a sprint (we are working our way to the point of treating advanced NSCLC as a chronic disease, but breast cancer is very typically comprised of many lines of therapy over years), how you do in the first few miles doesn’t necessarily have much to do with how to do at the end of the race.

What is more puzzling is the finding on the FLEX trial (cisplatin/navelbine with or without Erbitux (cetuximab)) of an improvement in overall survival without an improvement in progression-free survival. Nobody has been able to explain how that would occur.

>Jim: “Dr. West (aka Batman) has discussed this issue before: http://cancergrace.org/forums/index.php?topic=1030.0”

Thank you Jim, for the link. And pardon me, everyone, for confusing DR. West with his namesake, ADAM West (http://en.wikipedia.org/wiki/Adam_West) and for channeling the Riddle (http://en.wikipedia.org/wiki/Frank_Gorshin).

Jim’s referenced article thread also included a discussion on the topic, worthy of note:
http://cancergrace.org/cancer-101/2008/11/09/pfs-good-and-bad/

The article thread contains a logical proposition that I have trouble accepting:

> Dr. West: “So if new drug X improves survival by 20% compared with standard chemo, but later the two groups converge toward each other because effective second and third line treatments blunt the differences after first line, the overall survival difference won’t be significantly better. Or it could be that people can just as easily get drug X later, and the end up doing just as well.”

(I wish I could quote better in this section of GRACE.)

If the comparison of OS in studies of a new treatment (drug, radiation, whatever) includes subsequent treatments, how do the researchers normalize for differences in those subsequent treatments? In the hypothetical given above, that drug X improved survival (or, more specific to my question, improved progression-free survival) by 20% compared to the baseline, from time Y to time Z, that’s an improvement, and, depending upon the numbers, significant. If subsequent treatments “narrowed” or eliminated the advantages, I think the wrong approach is to just shrug one’s shoulders and say “ah, another failed study”. Rather, more research is needed to understand what it is about the subsequent treatments that eliminated the advantage early in the race.

For example:
* I’ve read that there’s a negative correlation between use of a Taxane in 1st line treatment and effectivity of Alimta in a later treatment (it was mentioned in one of Dr. West’s webinars with a guest speaker, among other things). If a large enough number of patients were treated with Alimta as 2nd or 3rd line treatment following 1st line Taxol, Taxol might have shown a lower OS than it otherwise could have. What’s a better answer - consider using an agent different than Alimta if the patient has been treated with a Taxane in 1st line therapy. However, had this “Alimta” effect been around (I don’t think it was, in terms of the timeline) while Taxanes were being studied, then, if I understand the method, the OS of Taxanes would have been reported as not as good - perhaps not good enough to warrant it as a standard therapy in some situations.

* The above example also makes me wonder if the OS differences in Abraxane (which, in theory, should be an outstanding addition to the NSCLC chemotherapeutic arsenal) aren’t diminished by this very effect. Moreover, if Breast Cancer research relies (more heavily? totally?) on PFS vice OS, no wonder Abraxane is approved for Breast Cancer, and not approved for NSCLC.

> Dr. West: “how you do in the first few miles doesn’t necessarily have much to do with how to do at the end of the race.”

See my comments, above. If what a group of runners do in the last few miles isn’t normalized, the benefit of a given technique or equipment in the first few miles might be lost. An in character example: imagine if I design a new wonderful shoe that mimics the human foot. I find it increases injury free sprints (IFS) and improves early times in a marathon by 20%. However, in a large, nationwide trial of marathon runners, the overall time is no better. So, by the logic presented in this article and those referenced, the Shoe & Sock Administration (SSA) disallows the new shoe, and credit card companies won’t let you buy them.

However, a stubborn researcher notes that the new shoe does poorly in long distances on hard or rough surfaces, and does great on smooth, soft, sandy or turf surfaces, where its advantage is maintained. New trials are performed, concentrating on the appropriate longer-term treatment of the surfaces. The advantage is maintained. The SSA approves the new shoes for marathons as well as sprints. The credit card companies allow their purchase for marathons as well as sprints. The runners have improved Quality of Life as they finish their sprints and marathons in less time and in more comfort.

I have to agree with Joe on this one. I have always had trouble understanding how good PFS could wind up being poor OS. Part of it I ‘m sure is how the numbers are interpreted and how researches are ‘forced’ to interpret. For instance, after a first line treatment of drug X which has a good or increased PFS, once progression occurs do all patients go on to the same second line treatment? If so that may be a slightly better indicator of OS but if they receive varying drugs then I’m not sure how the OS could really be accurate. Seems to me the only reliable number is the PFS because on a line of treatment the endpoint of PFS can be reached without the introduction of variables that can skew the numbers. I know science doesn’t like to assume anything but it would seem that PFS is a more important endpoint for studies and it could be ‘assumed’ that an increased PFS for a large percentage of patients in a study could translate to longer OS. What becomes important then is individualized treatment that can continue the good results of a drug that offers a good PFS. Sorry if this seems simplistic but when it comes to stats and their interpretations the results deserve multiple analyses. –mikem

I must admit, I do not know much about the difference between PFS and OS except that at some level I do understand. Just wanted to say thanks for this post which does such a good job of asking so many questions lung cancer patients have on a daily basis. Hopefully through endeavors such as this post, we can begin to answer the questions and achieve some kind of equality in the “race for the cure”. To continue the runner analogy, it would be nice to at least be recognized as being a participant in the race.

I think my point is that if you aren’t controlling for what happens after the duration of the study, and everyone comes out with the same survival, I would argue that the magnitude of the benefit just isn’t very great. This isn’t to say that there is NO value in a prolonged progression-free survival, but rather that we simply can’t pay for every treatment to be given to every patient no matter how questionable the benefit is. These interventions aren’t an aspirin per day. Instead, they’re medicines that cost hundreds of millions or more than a billion dollars to the health care system, and as we tell people who have inadequate insurance that we can’t provide clearly life-prolonging, or sometimes even potentially curative therapies because there isn’t enough money out there, I think it’s appropriate to take a hard look at what you get for the wild costs of many of these newer drugs.

Parenthetically, I’d say that the evidence that there is a significant difference in how patients do on Alimta after a taxane versus a different chemotherapy is the tiniest wisp of real data, just a vague hint in a subset analysis of a single trial once. And we know that if you do 20 subset analyses, one will emerge as positive just by random chance (p value 0.05), which is part of why we try to restrain ourselves from drawing broad conclusions based on an endless array of subset analyses that are, in many cases, not much more than a fishing expedition. In the absence of corroborating evidence, I’d say that there is almost no traction for this vague hypothesis of taxanes leading to lower efficacy with subsequent Alimta just yet. If the maintenance trial with Alimta were to show that finding, I’d consider it far more likely to be real than just a random finding that is the byproduct of doing too many analyses. As they say, “if you torture the data enough, they’ll tell you anything you want.”

Taking a step back, I think we need to realize that we’ve all got our biases here and that, to people who are managing cancer on a personal basis in themselves or an immediate family member, there may be no amount of money for which there is inadequate value in a potential treatment. But the fact is that our resources are a zero-sum game. If our system provided full finding for treatments that were significantly life-prolonging for everyone but then allowed people to buy their own additional interventions with perhaps a 50% co-pay, what would people find “worth it”?

We’re at a point where we can’t pay for every remotely arguably helpful medical intervention without actually making a choice of what we’re going to deny to someone else. So the question isn’t “is this better than nothing?”, but rather, “is this better than pre-natal care for women in desperate poverty?” or some other program we’re cutting because of the skyrocketing costs of medical care.