Cancer Study Headline, Translated: When RECIST and PFS Show Up

Two documents, one word

Two very different pieces of paper use the word "response." One is a trial press release. The other is the scan report in your own chart. RECIST is the rulebook that links them, and knowing the rules changes how both read.

RECIST stands for Response Evaluation Criteria in Solid Tumors. The National Cancer Institute describes it as a standard way to measure how well a patient responds to treatment, based on whether tumors shrink, stay the same, or get bigger. To use it, there must be at least one tumor that can be measured on an X-ray, CT, or MRI.

That last clause matters. RECIST is for solid tumors that can be seen and measured. It does not apply to leukemia, and it handles bone-only disease and fluid collections awkwardly.

How tumors become one number

A radiologist does not measure everything. The rules pick a sample.

Up to 5 target lesions are chosen in total, with a maximum of 2 per organ. Paired organs count as one organ. Lung, kidney, ovaries, and lymph nodes each count once, no matter how many separate parts they have.

Each target lesion is measured in one dimension, along its longest diameter. Lymph nodes are the exception. A node counts as measurable only if its short axis is 15 millimeters or more.

Everything else is recorded as a non-target lesion. Those are watched, but not measured with a ruler.

Then all the target diameters are added into a single sum. That sum is the number the whole system runs on.

The four categories, and their real thresholds

• Complete response, or CR. All target lesions disappear.
• Partial response, or PR. The sum of diameters falls by 30 percent or more.
• Progressive disease, or PD. The sum rises by 20 percent or more, with a minimum absolute increase of 5 millimeters. New lesions also count as progression on their own.
• Stable disease, or SD. Anything that is neither PR nor PD.

Two things follow from those numbers. A tumor sum that drops 25 percent is officially stable disease, not a response. And a small tumor can cross the 20 percent line with a few millimeters of growth, which is exactly why the 5-millimeter floor exists.

Our page on complete and partial response covers what these labels mean outside a trial, and what RECIST means walks through the terms.

The nadir, which nobody explains

Progression is not measured from where you started. It is measured from the smallest sum recorded so far, called the nadir.

The RECIST Working Group gives a worked example. Say the sum of longest diameters is 165 millimeters at baseline, 63 millimeters at week 8, and 65 millimeters at week 16. Week 16 is still a continuing partial response. The shrinkage is judged against baseline. The growth is judged against the nadir.

That asymmetry is deliberate. It catches regrowth early without punishing a treatment for a couple of millimeters of measurement noise.

Where PFS comes from

Progression-free survival is the length of time a person lives with the disease without it getting worse. In a trial, "getting worse" almost always means RECIST progression on a scheduled scan.

So PFS is not a continuous measurement of a person's condition. It is a stopwatch that stops at a scan appointment. If one arm of a trial is scanned every 6 weeks and the other every 12, the same biology produces different PFS numbers. This is one reason PFS results are argued over more than survival results.

Immunotherapy broke the rules

Immunotherapy drugs sometimes cause tumors to swell before they shrink, or produce new spots that later disappear. This is called pseudoprogression. Under plain RECIST 1.1, that looks like progression, and a working treatment could be stopped.

iRECIST was written to handle it. It adds unconfirmed progression and confirmed progression, and it requires a repeat scan to verify progression before therapy is stopped.

A worked example: pancreatic cancer

Pancreatic cancer sits behind many PFS headlines, so it is a useful case.

NCI names jaundice, pain, and weight loss as the signs, and states plainly that this cancer is difficult to diagnose early. Smoking and health history affect risk. Tests that examine the pancreas are used to diagnose and to stage it.

Surgery is the only route to cure, and only some tumors allow it. NCI describes the Whipple procedure, which removes the head of the pancreas, the gallbladder, part of the stomach, part of the small intestine, and the bile duct. Other options are total pancreatectomy and distal pancreatectomy. When the cancer cannot be removed, palliative surgery such as a biliary bypass or an endoscopic stent can relieve a blocked bile duct.

When to get checked

There is no routine screening for pancreatic cancer in people at average risk. Symptoms carry the weight. See a clinician promptly for:

• Yellowing of the skin or the whites of the eyes.
• Dark urine with pale, greasy stools.
• Itching without a rash.
• Pain in the upper abdomen that goes through to the back.
• Weight loss you did not intend.
• New diabetes after age 50, especially with weight loss.

The numbers a PFS gain sits against

These SEER figures describe a large population, not a person.

Five-year relative survival for pancreatic cancer is 13.7 percent for cases from 2016 to 2022. By stage it is 43.6 percent while confined to the pancreas, 17.0 percent once it reaches nearby lymph nodes, and 3.4 percent once it has spread. Only 15 percent are found at the first stage, and 51 percent are already distant at diagnosis. Median age at diagnosis is 71.

Against numbers like those, a trial reporting a two-month PFS gain is neither trivial nor a cure. Knowing exactly what was measured is the only way to judge which. Our guides to what clinical trials are and joining a clinical trial cover the rest.

Sources

• NCI Dictionary of Cancer Terms — RECIST
• NCI Dictionary of Cancer Terms — Progression-free survival
• RECIST Working Group — RECIST 1.1
• RECIST Working Group — Home and FAQ, including iRECIST
• NCI PDQ — Pancreatic Cancer Treatment (Patient Version)
• NCI SEER Cancer Stat Facts — Pancreatic Cancer