Does Low-Dose Radiation Help Arthritic Knees, or Do Patients Just Feel Better?

For two weeks at Sint Maartenskliniek in Nijmegen, some patients with painful knees went through six radiotherapy sessions. Roughly half of them received no beam. In the Dutch sham-controlled trial, the control group heard a recorded playback of a radiotherapy session in place of treatment, and the other group received six fractions of 1 Gy, for a total of 6 Gy.

At three months, 44% of the irradiated patients and 43% of the sham patients met standard responder criteria (p = 0.9). One percentage point separated the real treatment from the recording.

The treatment being tested, low-dose radiation therapy (LDRT) for osteoarthritis, is not new. It is used regularly in Germany and Spain and widely in German-speaking countries. The question these trials asked was simple to state and hard to answer: when patients feel better after low-dose irradiation of an arthritic knee, how much of that improvement comes from the radiation?

The answer matters because a therapy can stay in routine use for years on the strength of patients who improve, and improvement by itself proves little. Only a comparison against an identical ritual without the dose can separate the beam from everything around it. At least four sham-controlled trials now exist: the Dutch knee and hand trials, an Iranian trial, and a Korean trial whose full results appeared on October 2, 2026.

Six sessions, one recording, two groups

The Dutch knee trial screened and randomized patients from October 14, 2015, to May 3, 2017. Eligible patients were at least 50 years old, rated their pain at 5 or more out of 10, and had not responded to analgesics and exercise therapy. Twenty-seven received LDRT and 28 received the sham.

A fraction is one session’s share of the total dose. Splitting 6 Gy into six 1 Gy fractions over two weeks matched the German DEGRO guidelines, as the sister trial in hand osteoarthritis noted. That trial used the same playback method.

The outcome measure was the OMERACT-OARSI responder criteria, a standardized definition of meaningful improvement in osteoarthritis pain and function. Under that definition the two groups were indistinguishable. In their Annals of the Rheumatic Diseases paper, published online in 2018 and in print in January 2019, Mahler and colleagues wrote that they “advise against the use of LDRT as treatment for knee OA.”

Follow-up did not change the picture. At six months, nine of 22 irradiated patients (41%) and nine of 26 sham patients (35%) were responders, with no significant differences at 6 or 12 months. The authors wrote that the large effects seen in practice “can probably be explained by a regression to the mean effect and response to placebo.”

Why a sham group improves

Regression to the mean is a statistical effect. People tend to seek treatment, and to enter trials, when their symptoms are at a peak, and symptoms that fluctuate will on average drift back toward a more typical level whether or not anything is done.

Placebo effects are another possible contributor. The improvement seen in a sham group is simply improvement without irradiation, and it may include both placebo effects and ordinary symptom fluctuation. The sham comparison therefore estimates what radiation adds beyond that improvement, rather than isolating the effect of receiving care. The Dutch recording was designed to mimic a real session.

The Dutch sham group’s 43% response at three months has a close counterpart in Korea, where the sham group’s responder rate was 41.7%. The Korean trial’s principal investigator, Dr. Byoung Hyuck Kim, a radiation oncologist at Seoul National University, said the high sham response was consistent with other osteoarthritis trials.

The Korean trial and its 3 Gy arm

The LoRD-KNeA trial took a different approach to dose. Its 2023 protocol specified six fractions of 0.5 Gy, for 3 Gy total, delivered with 6 MeV linear-accelerator beams, and it required radiographic Kellgren-Lawrence grade 2 or 3 disease. A third arm received 0.3 Gy in six fractions.

The full trial report in BMC Medicine describes 114 patients aged 50 or older with walking pain of 50 to 90 on a 100-point scale. Sham patients went through the same setup with no beam delivered, and only acetaminophen was allowed for the first four months, according to the trial’s ASTRO release.

At four months, in the modified intention-to-treat analysis, 70.3% of the 3 Gy group responded, against 41.7% of the sham group (p = 0.021). The 0.3 Gy group reached 58.3%, which was not significantly different from sham (p = 0.168). When the short-term results were presented at ASTRO on September 28, 2025, the abstract reported p-values of 0.014 and 0.157, so the full paper’s figures are somewhat weaker.

Across the two knee trials, then, most of the improvement after low-dose irradiation appears in the sham groups too. The Dutch trial at 6 Gy found no difference from sham, and the Korean trial shows a modest signal at 3 Gy on a single outcome.

How far the signal reaches

The positive Korean result rests on the responder rate. The ASTRO release reported a WOMAC response of 56.8% versus 30.6% (p = 0.024), but in the full paper, mean changes in WOMAC subscales, visual analog pain scores, patient global assessment, inflammatory markers, and MRI measures did not differ significantly among groups.

Durability was limited too. The 3 Gy advantage persisted at eight months and was no longer statistically significant at 12.

No treatment-related toxicities were reported. That is a short-term observation and says nothing about late effects.

Elsewhere, a 60-patient Iranian trial comparing 3 Gy with sham reported improvement, as summarized in a 2025 review. The 229-patient ArthroRad trial compared 3.0 Gy with 0.3 Gy and found pain improved in both arms. It had no sham arm, so it cannot separate either dose from placebo. A 2025 meta-analysis, which predated the full Korean paper, found no significant benefit over sham for pain or function and reported mild adverse effects.

An argument about dose per fraction

The trials are not easy to set side by side. They differ in eligibility, analgesic rules, dose per fraction, and timing of the primary endpoint, at three months in the Netherlands and four in Korea. They are also small. The Dutch hand trial was powered to detect a 40% difference, which means it cannot rule out smaller effects.

Dose per fraction may matter. A 2022 review wrote that newer data “may suggest more anti-inflammatory response with 0.5 Gy per fraction for a total of 3 Gy.” That remains a hypothesis. The authors of the German KORT protocol, published February 27, 2026, describe earlier sham trials as using “non-standard dose concepts”, a reading that sits awkwardly beside the Dutch trial’s adherence to DEGRO guidance.

The next tests are already designed around a sham. KORT is sham-controlled, and the registered Radio-KO trial compares six 0.5 Gy fractions with a sham, with a second course for non-responders at two months.

Each will need a control group that goes through every step of the treatment except the dose. In Nijmegen, that group ended up one percentage point from the irradiated patients.