What Is a Deauville Score and What Do the Results Mean?

A Deauville score is a five-point rating system doctors use to interpret PET/CT scans in people being treated for lymphoma. It works by comparing how much radioactive glucose (FDG) a suspicious area absorbs relative to two internal reference points in the body: the mediastinum (the area between the lungs containing blood vessels and the heart) and the liver. A score of 1, 2, or 3 generally signals a good response to treatment, while a 4 or 5 suggests the disease may still be active. The system is now embedded in the international Lugano classification, the standard framework for staging lymphoma and evaluating whether treatment is working.

The Five-Point Scale

Every Deauville score compares FDG uptake at a site of known or suspected lymphoma to what the mediastinum and liver absorb on the same scan. The liver absorbs more FDG than the mediastinum under normal circumstances, so the scale creates a staircase of increasing concern:

Score 1: No uptake at all at the disease site.
Score 2: Uptake at the site is present but equal to or less than the mediastinum.
Score 3: Uptake is higher than the mediastinum but equal to or less than the liver.
Score 4: Uptake is moderately higher than the liver.
Score 5: Uptake is markedly higher than the liver, or new lesions have appeared.

Some reporting systems also include a category “X,” which flags new areas of uptake that are probably unrelated to the lymphoma, such as infection or inflammation elsewhere in the body.¹ In practice, the critical dividing line falls between scores 3 and 4. Scores of 1 through 3 are broadly interpreted as a negative (favorable) PET result, meaning the treatment appears to be working. Scores of 4 and 5 are interpreted as positive (unfavorable), raising the question of whether the cancer is still metabolically active.

Why the Mediastinum and Liver Are the Yardsticks

PET scans measure how greedily cells consume glucose, and cancer cells tend to be heavy consumers. But the raw brightness of a spot on a PET image means nothing on its own because overall FDG uptake varies from patient to patient and even scan to scan, depending on blood sugar levels, how long the tracer circulated, and scanner settings. The Deauville system sidesteps this problem by using two tissues inside the patient’s own body as built-in benchmarks. The mediastinal blood pool sets the low bar, and the liver sets the higher bar. As long as those two organs behave normally, every patient carries their own internal ruler.

That design is elegant but has a known vulnerability. Fatty liver disease lowers the liver’s FDG uptake, which shrinks the gap between the liver benchmark and a residual lymphoma lesion. In patients with steatosis, a study found that average liver uptake values were meaningfully lower than in patients without fatty liver. After applying a correction factor to account for the steatosis, some patients who had been scored a 4 on interim scans were reclassified as 3, flipping them from a “positive” to a “negative” result.² Since fatty liver is common, this is not a trivial edge case. If your scan report says Deauville 4 and you also happen to have fatty liver, your oncologist may factor that in before concluding the treatment is failing.

Interim Scans Versus End-of-Treatment Scans

The Deauville score is applied at two major decision points during lymphoma treatment. The first is the interim scan, typically done after two cycles of chemotherapy. The second is the end-of-treatment scan, performed after the full course of therapy is complete. The two scans serve different purposes.

An interim scan is essentially a progress check. If the disease is responding well (scores 1 through 3), some treatment protocols allow doctors to consider de-escalating therapy to reduce side effects. If the scan is positive (scores 4 or 5), the team may switch to a more aggressive regimen. A Cochrane systematic review found that interim PET results evaluated with the Deauville scale were the dominant method for predicting outcomes in Hodgkin lymphoma across multiple studies.³ For diffuse large B-cell lymphoma (DLBCL), the most common aggressive form of non-Hodgkin lymphoma, interim Deauville scoring also proved to be among the most accurate methods for predicting end-of-treatment response, with an accuracy around 82%.⁴

An end-of-treatment scan, on the other hand, provides the verdict. A score of 1 through 3 at this stage means the patient has achieved what is called a complete metabolic response. A score of 4 or 5 suggests residual active disease and usually triggers further workup, additional treatment, or closer surveillance.

What Different Scores Mean for Prognosis

The prognostic weight of a Deauville score varies depending on the type of lymphoma, the timing of the scan, and the rest of the clinical picture. But the broad pattern holds across several lymphoma subtypes: patients with scores of 1 through 3 consistently fare better than those with scores of 4 or 5.

In DLBCL, a multi-institutional study found that patients with end-of-treatment scores of 1 through 3 had a five-year recurrence-free survival rate around 87%, compared with roughly 67% for those scoring 4 or 5. The gap in overall survival was even starker, at about 87% versus 62%.⁵ And the score’s predictive power becomes even stronger when combined with other clinical information. Research on DLBCL patients found that combining interim Deauville scoring with a baseline prognostic index created a three-tier risk stratification. Patients in the high-risk group (scores 4–5 plus a high baseline risk index) had roughly fourteen times the hazard of disease progression or death compared to the low-risk group.⁶

In follicular lymphoma, a slower-growing type, the picture has historically been murkier because follicular lymphoma cells do not always light up strongly on PET scans. A meta-analysis found that a positive interim PET was associated with nearly triple the risk of disease progression and roughly triple the risk of dying, confirming that the Deauville score does carry prognostic value in this setting as well.⁷ End-of-treatment positivity was also found to be an independent predictor of progression.⁸

How Scores Guide Treatment Changes

A favorable interim Deauville score does not just offer peace of mind; in some lymphoma types it actively changes the treatment plan. In advanced-stage Hodgkin lymphoma, for instance, clinical trials have shown that patients with a negative interim PET after initial chemotherapy can safely be stepped down to a less toxic regimen, sparing them side effects without sacrificing cure rates. Meanwhile, patients with a positive scan can be escalated to a more intensive chemotherapy combination.

There is a limit to that strategy, though. For patients scoring 5 after two cycles of standard ABVD chemotherapy, a pooled analysis of three multicenter trials found that treatment intensification did not rescue outcomes well. This high-risk subgroup had an inadequate response to escalation, suggesting that a score of 5 on an interim scan may identify patients who need altogether different approaches, such as novel agents or clinical trials, rather than just a stronger version of conventional chemotherapy.⁹

In DLBCL, end-of-treatment Deauville scores have been used to decide whether to add consolidative radiation therapy. A study found that radiation significantly improved time to progression for patients with scores of 4 or 5 but made no meaningful difference for patients with scores of 1 through 3, where the disease was already in good metabolic remission.¹⁰ The score essentially helps answer the question: does this patient need more treatment, or can we stop?

One important caveat: some early validation studies were conducted without altering treatment based on interim scan results, meaning the prognostic value was confirmed in settings where the scan was read but not acted upon.¹¹ The studies showing that it is safe to change treatment based on these scores came later and are now well established for Hodgkin lymphoma, with growing evidence in DLBCL.

False Positives and the Score 4 Gray Zone

One of the most anxiety-producing situations for patients is receiving a score of 4 or 5 when the disease is actually gone. False positives happen. Inflammation from treatment, healing scar tissue, infections, and even granulomas (small clumps of immune cells unrelated to cancer) can all light up on a PET scan. A pathologic validation study biopsied lesions in patients with various Deauville scores after treatment and found that among score-4 lesions, about 30% were not lymphoma. Among score-5 lesions with new areas of uptake, roughly a quarter were false positives. Meanwhile, the false-negative rate among score 1-3 lesions was very low, with only about 8% of those cases having lymphoma on follow-up.¹² Non-lymphoma findings included granulomas, non-specific inflammation, other types of cancer entirely, and areas with no residual disease at all.

This is why a Deauville score of 4 is often treated as a gray zone. It falls above the liver threshold but could represent active lymphoma, residual inflammation, or something else entirely. In clinical practice, a score of 4 does not automatically trigger a treatment change. Many oncologists will want a biopsy to confirm disease before escalating, or they may adopt a wait-and-rescan approach. The score is a tool for decision-making, not a final answer.

How Consistently Do Doctors Agree on Scores?

A common concern, especially in clinical trials where treatment decisions hinge on a specific score, is whether two different readers looking at the same scan would assign the same number. The answer is: usually, but not always, and the middle of the scale is where disagreement concentrates.

In Hodgkin lymphoma, a study of inter-reader reliability after two cycles of chemotherapy found that the probability of two random readers assigning exactly the same score was only 42%. Agreement was highest at the extremes (scores 1 and 5, each above 50%) and lowest for score 2 (25%) and score 3 (36%). However, when the scale was collapsed to a simple binary question — is this scan positive or negative? — agreement jumped to 86% when the cutoff was placed between scores 3 and 4.¹³

In DLBCL, agreement was generally higher. Overall agreement was around 88% for interim scans and 92% for end-of-treatment scans. But agreement on positive cases specifically was lower (about 74–76%), suggesting that when a scan looks abnormal, there is more room for disagreement about exactly how abnormal it is.¹⁴ For this reason, clinical trials increasingly use central review panels where multiple experts independently score each scan and a consensus is reached. For routine clinical care, the binary positive-or-negative interpretation is usually enough, and the agreement there is strong.

Technical Factors That Can Shift a Score

Beyond fatty liver, several other technical and biological factors can influence the number on your report. Growth factor injections (G-CSF), commonly given to patients during chemotherapy to boost white blood cell counts, can dramatically increase FDG uptake in bone marrow and spleen. This extra background noise can make it harder to compare disease sites to the liver benchmark and may affect the assigned score. Research has investigated the optimal timing between the last G-CSF injection and a PET scan to minimize this interference.¹⁵

Even the way the PET scanner reconstructs its images can matter. A study comparing two different reconstruction methods found that the agreement between resulting Deauville scores was high but not perfect, with a small fraction of patients receiving a different score depending on the algorithm used.¹⁶ Standardizing reconstruction protocols across centers has become a priority in multi-site clinical trials for exactly this reason.

Pediatric Lymphoma and the Limits of the Score

The Deauville score was developed and validated primarily in adult lymphoma patients, and its performance in children does not always match. One study of pediatric Hodgkin lymphoma found that using Deauville criteria at the end of treatment flagged very few patients as positive — only about 6% — and of those, two-thirds turned out to be false positives.¹⁷ Children’s treatment protocols often differ from adults’, and the biology of their tumors can behave differently after therapy. The Deauville score remains part of the evaluation toolkit in pediatric oncology, but it is interpreted more cautiously and in the context of a broader clinical picture.

Use Beyond Lymphoma

Although the Deauville score was designed for lymphoma, researchers have explored whether the same framework can be applied to other cancers that show up on PET scans. The most developed application outside lymphoma is in multiple myeloma, a cancer of the bone marrow plasma cells. A study published in the Journal of Clinical Oncology applied the five-point Deauville scale to PET scans in newly diagnosed myeloma patients, scoring both bone marrow and focal bone lesions, and then tested whether those scores predicted outcomes.¹⁸ The Lugano classification that formalized Deauville scoring for lymphoma has encouraged standardization in PET interpretation more broadly, and myeloma is an area where that standardization is actively being tested.¹⁹

Quantitative Extensions and Automation

A longstanding frustration with the Deauville score is that it is inherently visual and subjective. Two readers looking at the same scan are comparing brightness by eye, which explains the agreement issues described earlier. To address this, researchers have developed a quantitative extension called qPET that translates the Deauville categories into a continuous numeric scale. Under this approach, Deauville 3 corresponds to a qPET value of about 0.95, Deauville 4 to about 1.3, and Deauville 5 to about 2.0, giving clinicians a more granular measure rather than forcing a lesion into one of five bins.²⁰

Artificial intelligence is also being applied to automate the reference-level measurements that underpin the score. An AI-based method for automatically segmenting the liver and mediastinal blood pool on CT images, and then transferring those regions onto PET images to calculate standardized uptake values, showed good agreement with measurements done manually by experienced radiologists.²¹ A systematic review of AI applications in oncologic PET imaging confirmed that these automated tools are performing comparably to human readers for the reference-region measurements.²² If widely adopted, such tools could reduce inter-observer disagreement by standardizing the part of the scoring process that relies on human judgment. The first- and second-line treatment evaluations in Hodgkin lymphoma appear to require the same response criteria, which simplifies both automated and human interpretation across treatment settings.²³