Progression-free survival

Progression-free survival (PFS) is "the length of time during and after the treatment of a disease, such as cancer, that a patient lives with the disease but it does not get worse".^[1] In Oncology, PFS usually refers to situations in which a tumor is present, as demonstrated by laboratory testing, radiologic testing, or clinically. Similarly, "disease-free survival" is when patients have had operations and are left with no detectable disease.

Time to progression (TTP) does not count patients who die from other causes but is otherwise a close equivalent to PFS (unless there are a large number of such events).^[2] The FDA gives separate definitions and prefers PFS.^[3]

PFS is widely used in oncology.^[4] Since (as already said) it only applies to patients with inoperable disease that are generally treated with drugs (chemotherapy, target therapies, etc.) it will mostly be considered in relation to drug treatment of cancer.

A very important aspect is the definition of "progression" since this generally involves imaging techniques (plain radiograms, CT scans, MRI, PET scans, ultrasounds) or other aspects: biochemical progression may be defined on the basis of an increase in a tumor marker (such as CA125 for epithelial ovarian cancer or PSA for prostate cancer). At present any change in the radiological aspect of a lesion is defined according to RECIST criteria. But progression may also be due to the appearance of a new lesion originating from the same tumor or to the appearance of new cancer in the same organ or in a different organ, or due to unequivocal progression in 'non-target' lesions - such as pleural effusions, ascites, leptomeningeal disease etc.

Progression-free survival is often used as an alternative to overall survival (OS): this is the most reliable endpoint in clinical studies, but it will only be available after a longer time than PFS. For this reason, especially when new drugs are tested, there is a pressure (that in some cases may be absolutely acceptable while in other cases may hide economical interests) to approve new drugs on the basis of PFS data rather than waiting for OS data.
PFS is considered as a "surrogate" of OS: in some cancers the two elements are strictly related, but in others they are not. Several agents that may prolong PFS do not prolong OS. PFS may be considered as an endpoint in itself (the FDA and EMEA consider it such) in situations where overall survival endpoints may be not feasible, and where progression is likely or very likely to be related to symptomatology.

Patient understanding of what prolongation of PFS means has not been evaluated robustly. In a time trade off study in renal cancer, physicians rated PFS the most important aspect of treatment, while for patients it fell below fatigue, hand foot syndrome, and other toxicities. <Park et al>

Special aspects of PFS

There is an element that makes PFS a questionable endpoint: by definition it refers to the date on which progression is detected, and this means that it depends on which date a radiological evaluation (in most cases) is performed. If for any reason a CT scan is postponed by one week (because the machine is out of order, or the patients feels too bad to go to the hospital) PFS is unduly prolonged.

On the other hand, PFS becomes more relevant than OS when in a randomized trial patients that progress while on treatment A are allowed to receive treatment B (these patients may "cross" from one arm of the study to the other). If treatment B is really more effective than treatment A it is probable that the OS of patients will be the same even if PFS may be very different. This happened for example in studies comparing tyrosine kinase inhibitors (TKI) to standard chemotherapy in patients with non-small cell lung cancer (NSCLC) harboring a mutation in EGF-receptor. Patients started on TKI had a much longer PFS, but since patients that started on chemotherapy were allowed to receive TKI on progression, OS was similar.

The relationship between PFS and OS is altered in any case in which a successive treatment may influence survival. Unfortunately this does not happen very often for second-line treatment of cancer, and even less so for successive treatments.

The advantage of measuring PFS over measuring OS is that PFS appears sooner than deaths, allowing faster trials and oncologists feel that PFS can give them a better idea of how the cancer is progressing during the course of treatment.

Traditionally, the U.S. Food and Drug Administration has required studies of OS rather than PFS to demonstrate that a drug is effective against cancer, but recently the FDA. has accepted PFS. The use of PFS for proof of effectiveness and regulatory approval is controversial.

It is often used as a clinical endpoint in randomized controlled trials for cancer therapies.^[5]

It is a metric frequently used by the UK National Institute for Health and Clinical Excellence^[6] and the U.S. Food and Drug Administration to evaluate the effectiveness of a cancer treatment. PFS has been postulated to be a better ("more pure") measure of efficacy in second-line clinical trials as it eliminates potential differential bias from prior or subsequent treatments.

However, PFS improvements do not always result in corresponding improvements in overall survival, and the control of the disease may come at the biological expense of side effects from the treatment itself.^[7] This has been described as an example of the McNamara fallacy.^[7][8]

References