In previous posts I have expressed my opinion that the spectacular pummeling of Sirtex followed by a fairly meteoric rise earlier this week was basically due to the fact that retail investors don’t understand the basics of clinical trials. I do think that Sirtex did a lousy job with communication (this week’s communication exercise was only marginally better) but there is no doubt that clinical trial design and outcomes are a complex matter and nobody should feel bad about not “getting it”.

Since my last post I received a dozen or so emails querying point #3 (survival benefit from SIRFLOX) and it is really clear from trading forums such as HotCopper that people don’t get what the ASCO abstract really means.

Additional 7.9 months! That is incredibly valuable to a person and their family. Not only will this be a headliner at the ASCO (in front of 30 000 oncologists), expect news to spread like wildfire. Buckle up! Risk is gone, growth is here

… and

In 2001 the pfs was 15.9 months, versus Hepatic artery chemo at 9.7 months. So an extra 6.2 months at that stage. This time we got an extra 7.9 months. Better, but 14 years later.

… and

Considering the huge increase in PFS in the liver over the past 10 years and all the benefits of new treatments, I’m more than happy that SIRT still gave an extra 7 months.

I note for full disclosure that I am not a clinical trialist or a biostatistican but in the interest of delivering the (notional) public benefit of this site, I am happy to exercise a little scientific journalism. I encourage readers to chime-in if I’ve missed something that should be discussed.

In my opinion there are four main issues that need to be understood about the clinical development of SIR-Spheres in the mCRC setting.

1) The overall goal of therapeutic development in the oncology setting

Despite wonderful progress on many fronts (as the feature image, borrowed from ASCO attests), we do not – with fairly few exceptions – “cure” cancer in the metastatic setting. We can really only hope that a therapeutic will do at least one of two things, namely prolong survival and/or improve the quality of life. Therapeutics that improve survival but put the patient through misery are certainly part and parcel of treating cancer. But increasingly, particularly with the advent of targeted molecular therapeutics, we have an array of drugs that are highly specific and well tolerated, and essentially put the cancer into “stasis”, at least for a period of time.

These two development objectives are a good point for understanding the difference in the clinical impact of overall survival (OS) and progression-free survival (PFS). A drug that confers OS extends patient lifespan by combating the disease – time is the measure of outcome. A drug that confers only PFS but not OS (for example, the outcome determined from the SIRFLOX study) does not extend patient life, but may improve the quality of life, or make the patient’s condition more manageable. How to incorporate it into patient care is therefore a much more complex discussion and is why Sirtex is (appropriately) not over-egging the pudding at this stage.

I will get onto that in a minute.

2) Why do we even care about PFS then?

This is a great question and it is still somewhat controversial in most oncology trial settings.

The reality is that measuring a survival outcome in late-stage cancer patients is tough for a whole variety of reasons but the main reason is that their anticipated lifespan is fairly short. You need to have a large number of patients (i.e. statistically significant) in order to measure a month or two of incremental improvement. Patients die from other causes. Patients drop out. Patients have complications related to their treatment. This is all factored in and therefore a grown-up study to really measure OS as a primary end-point is big (many sites), hard work and expensive.

As an aside note, a lot of people in the general public assume that getting a product “approved” (i.e. marketing clearance) by a regulatory agency means that everything is golden. Not true. It’s one thing to convince the FDA (or TGA) that a product is safe and has therapeutic efficacy (and yes, PFS can do that), it’s another thing to convince insurance companies and healthcare systems that using a particular therapeutic confers sufficient benefit to warrant the cost. That’s why OS is actually important – if you can show meaningful survival benefit it’s a ticket to print money.

PFS is kind of an odd beast in some ways. A decade ago we hardly ever saw PFS as a clinical trial outcome. Today it is quite common precisely because of the types of targeted drugs that are being developed. PFS is also, quite frankly, a byproduct of two major commercial drivers. Firstly, pharma companies are trying to run smaller, more cost-effective trials with their targeted drugs, in patient populations that are better defined. Because these drugs have a targeted mechanism of action, disease progression is actually a fairly reasonable way of understanding whether a drug works or not. The second driver, believe it or not, is just availability of patients. There are so many drugs being developed, if we couldn’t come up with ways to ascertain efficacy in a relatively short window of time, we wouldn’t be able to stretch the patient pool. Moreover, with a rapidly improving standard of care, you need to be able to quickly put a patient onto another regimen (i.e. the “standard”) if you can’t ascertain improvement. To do otherwise is basically unethical.

By the way, this is the reason why SIRFLOX is used in conjunction with FOLFOX6m/Bevacizumab. It’s about making sure that the patients get access to the current best standard of care.

3) So really, SIR-Spheres are about managing progression of a disease?

The short answer is yes. 

SIR-Spheres are really not a curative therapeutic. In some cases, patients may ultimately end up with a more resectable disease (i.e. can have surgery that they would not otherwise be able to have) and this could have a significant survival benefit. But the real benefit of SIR-Sphere technology is that it may buy patients a period of relative stability even if it doesn’t ultimately extend life. Most mCRC patients die of liver failure so if you can delay that event and keep a patient relatively healthy, that’s a good thing for the patient and good for the hospital system. There is also no doubt that managing hepatic (liver) stability while undergoing a pretty aggressive therapy regimen (like FOLFOX6) is a good idea too.

4) But isn’t PFS really just a faster, easier predictor of OS?


PFS does not predict OS. 

PFS does not “hint” at OS. 

OS can only be measured when a trial, statistically powered to measure OS, is complete. Sirtex has been very clear about this and, in my view, communicated this correctly. When both the SIRFLOX and FOXFIRE studies are complete, there will be sufficient data to understand whether there is an improved survival benefit in conjunction with the standard of care. Not before.

To be absolutely clear, there has been no robust scientific study undertaken in any cancer setting to show that PFS predicts OS. Because there have been more drug trials in mCRC and ovarian cancer than many other diseases, there is some meta-analysis that suggests PFS may have some predictive value in treating systemic disease in certain patient subsets, but certainly not in an organ-specific context like SIR-Spheres.

A few concluding remarks

I really like Sirtex and I appreciate what the company is trying to do. I think the clinical argument is compelling and it is going to be fascinating to see how the final data stacks up. If the clinical discussion around the role of SIR-Spheres in patient management comes out positive, this could be a very successful product in the mCRC setting. You have probably gathered that I have some doubts that it is going to be as cut-and-dry as that, but we shall see.

I think SIR-Spheres will have a couple of saving graces that could be interesting for the future of the company. The first is that it is possible to image patients and confirm localisation of the SIR-Spheres. This is very significant and, in my opinion (remember I am not an oncologist), under-appreciated aspect of the product. This means that in the analysis of mCRC data, there may be a subset of patients that do predictably well on the basis of the biodistribution of SIR-Spheres in the liver, including suitability for resection. I note, incidentally, that OncoSil will never be able to do this (P32 is a pure beta emitter). Imaging-based biomarkers could have the same impact on the use of SIR-Spheres as the detection of KRAS mutations had on the use of cetuximab and panitumumab.

The second possibility is the combination of SIR-Spheres and cancer immunotherapy strategies like checkpoint inhibitors (i.e.PD-1/L1). Not much is more immunogenic than radiation and there are a lot of very interesting drugs coming down the path that are going to potentially shift the standard of care in the metastatic setting to a whole different level. I’m not actually that negative about the future of SIR-Spheres against the backdrop of cancer immunotherapy (many are). Indeed, radiation-based strategies may have more, not less, importance in this new world.

But in terms of any potential for an OS benefit relative to the current standard of care, I am doubtful. We have to remember that regorafenib and ziv-aflibercept (as systemic therapies) conferred less than 1.5 months of median survival benefit relative to “chemo”. This is a very high bar and shareholders set their expectations appropriately.

3 thoughts on “SIRFLOX : OS v PFS

  1. This is a really great post. You are correct I think in assuming that the market has at least partially misunderstood these two releases. If someone is to invest in these companies, they really need to understand what is being tested in the trial and what the potential outcomes mean for the company. SRX could have done a better job in explaining these results too.



  2. Another reason why PFS was an appropriate end point for the SIRFLOX trial is that they were looking at the application of SIRT as a first line therapy. Generally patients don’t die when their first line therapy fails, they progress and then get treated with second line, third line etc. This can make it tricky to evaluate what the impact of the first line therapy was on OS as different patients may end up getting different second and third line therapies. However, still not quite sure why the company, when the FDA stipulated choosing between two previous co-primaries (overall PFS and liver-specific PFS) they went for overall PFS. Strange choice given that SIRT is, by definition, a localised treatment, why would you then measure the effectiveness of your therapy by looking everywhere in the body.


Say something useful (or at least interesting)...

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s