How Tumor Response Is Measured and Where RECIST Falls Short

The short answer

Tumor response is measured by tracking a handful of representative lesions on serial CT or MRI scans and comparing their combined size against the baseline scan. The most widely used ruleset, the Response Evaluation Criteria in Solid Tumors (RECIST 1.1), converts those measurements into four categories: complete response, partial response, stable disease, and progressive disease. A roughly 30 percent shrinkage in the summed lesion diameters counts as a partial response, and a roughly 20 percent increase counts as progression. This framework works well for chemotherapy, but immunotherapy produces growth-then-shrinkage patterns that the original rules can misread, which is why a modified version called iRECIST now exists to require confirmation before calling a treatment a failure.

How RECIST 1.1 actually counts

RECIST 1.1, described by Eisenhauer and colleagues in 2009, imposes a deliberately rigid measurement discipline so that different radiologists at different centers arrive at the same answer. At baseline, the reader selects up to five measurable lesions total, with no more than two per organ, choosing the largest and most reproducibly measurable ones as target lesions. Each solid lesion is measured along its longest diameter, while enlarged lymph nodes are measured along their short axis. Those numbers are added together into a single sum of diameters, which becomes the reference point for every future scan.

From there the arithmetic is straightforward. A complete response means every target lesion disappears and pathological nodes shrink below 10 mm on the short axis. A partial response requires at least a 30 percent decrease in the sum of diameters. Progressive disease is defined as at least a 20 percent increase in that sum relative to the smallest value recorded during the study, with an added absolute requirement of at least a 5 mm increase to prevent tiny measurement noise from triggering a false alarm. Anything in between counts as stable disease. Lesions that are tracked but not measured, along with any brand-new lesion, are handled separately, and the appearance of an unequivocal new lesion signals progression on its own.

The strength of this system is consistency. By forcing readers to pick lesions in advance and follow the same ones over time, RECIST reduces the temptation to cherry-pick whichever lesion supports a desired conclusion. That reproducibility is why regulators and clinical trials lean on it as a primary endpoint.

Where the rules strain

The same rigidity that makes RECIST reliable also makes it a blunt instrument. It reduces a complex, heterogeneous disease to a one-dimensional length measurement of a few chosen lesions. A tumor can become more aggressive biologically without crossing the 20 percent threshold, and a lesion can swell from inflammation, hemorrhage, or scar tissue without containing more viable cancer. Size is a proxy for tumor burden, and the proxy is imperfect.

RECIST also treats every patient's tumor as though it responds as one unit. In reality, one lesion may shrink while another grows, a pattern that the summed-diameter approach can average into a misleadingly neutral verdict. And because the criteria were built around drugs that kill dividing cells, they assume that a shrinking tumor means the drug is working and a growing tumor means it is not. Immunotherapy violates that assumption.

Why immunotherapy needed a rewrite

Immune checkpoint inhibitors do not attack the tumor directly. They release the brakes on a patient's own T cells, which then infiltrate the tumor. That influx of immune cells can transiently enlarge a lesion, or even make previously invisible micrometastases briefly detectable, before the tumor begins to regress. Under strict RECIST 1.1, that early swelling looks like a 20 percent increase, gets labeled progressive disease, and can lead a clinician to stop a therapy that was on the verge of working.

This phenomenon is called pseudoprogression: an initial apparent increase in tumor burden followed by genuine shrinkage. It is real but uncommon. As reviewed in the Journal of Thoracic Disease, pseudoprogression rates with checkpoint inhibitors generally do not exceed roughly 10 percent, and in many tumor types they are lower. The clinical stakes are asymmetric, though, because wrongly discontinuing an effective drug in that minority carries a real cost.

The opposite failure mode is hyperprogression, a paradoxical acceleration of tumor growth after starting immunotherapy. Reported incidence has ranged widely across studies, from about 4 percent to 29 percent, in part because definitions differ. One common approach, described by Champiat and colleagues, defines it as at least a doubling of the tumor growth rate compared with the pre-treatment trajectory, which requires a scan from before therapy began to establish the baseline slope. RECIST 1.1, which only compares against the on-treatment baseline, was never designed to detect a change in growth rate.

What iRECIST adds

To handle these atypical patterns, the RECIST working group published iRECIST in Lancet Oncology in 2017, led by Seymour and colleagues. It keeps the RECIST 1.1 measurement machinery intact but changes how progression is confirmed. When a scan first meets the RECIST 1.1 definition of progression, iRECIST labels it immune unconfirmed progressive disease (iUPD) rather than declaring failure outright. A confirmatory scan is then performed no sooner than 4 weeks and no later than 8 weeks afterward.

If that follow-up scan shows further growth, the result becomes immune confirmed progressive disease (iCPD), which requires evidence such as a continued increase in the target lesion sum of at least 5 mm, worsening non-target disease, or enlargement of new lesions. If instead the tumor has shrunk or stabilized, the progression clock effectively resets and treatment can continue, capturing the delayed responders that strict RECIST would have discarded. New lesions, which end the assessment under RECIST 1.1, do not by themselves confirm progression under iRECIST; they trigger iUPD and must worsen at the next scan to count.

The scope of iRECIST is narrower than it might appear. It was designed primarily for consistent data collection in immunotherapy trials, and the RECIST group still recommends RECIST 1.1 for primary endpoints in phase 3 studies, with iRECIST as a structured exploratory layer. It reduces the risk of prematurely abandoning a working drug, but it does not solve hyperprogression, which needs pre-treatment imaging that response criteria alone cannot supply. As reviewed in the World Journal of Clinical Oncology, none of these frameworks measures viable tumor directly; they infer biology from geometry.

The broader lesson for reading oncology evidence is to ask which criteria a trial used, how progression was confirmed, and whether the response pattern fits the drug's mechanism. A number labeled progressive disease under one ruleset may read very differently under another.

This article is educational and is not medical advice.