Low-Dose Radiotherapy for Knee Osteoarthritis

Osteoarthritis (OA) of the knee is incredibly common and one of the main indications that radiation oncologists see when they start treating benign conditions. For anyone who’s treated it, it’s completely clear that it works, but it’s different when you’re trying to justify it to an insurance company or encourage uptake in a healthcare system rather than on an individual patient basis.

So the question is: Is there enough evidence to support this treatment in a widespread way?

First the non-randomised evidence

We know from large case series, both retrospective and prospective, totalling >7000 patients that you get a 70–80 % response rate in painful osteoarthritis across various joints. I’ve taken much of the info in the table from the DEGRO guidelines, but have added a weighted average response rate at the end. It may seem there is variability in response rates between different joints, but that may well be artefactual, as your response rate will depend on OA grading, duration of pain and other factors.

So that’s “interesting”, but in this post I’ll focus specifically on the randomised controlled evidence for knee osteoarthritis as there’s a lot of very recent (and some “less accessible”) data that’s not been incorporated into the recent reviews on the subject.

But first… why the knee?

Well, firstly because this is where the data is. But this begs the next question, which is why are people doing OA LDRT trials in knee in particular?

First, knee OA is very common. Second, it’s a relatively simple joint from both an anatomical and a radiotherapy-planning perspective.

But there is some complexity, which may well muddy the waters somewhat. In particular, the knee isn’t just one compartment – there’s the tibiofemoral joint (medial and lateral) and the patellofemoral joint. Many studies don’t clearly specify which compartments are involved, so results often mix patellofemoral-dominant and tibiofemoral-dominant disease. That’s worth keeping in mind when interpreting outcomes.

Also, since the knee is a weight-bearing joint, perhaps BMI (as well as metabolic status) could affect response rate, which we should bear in mind when using data from healthier populations (e.g. South Korea) to justify treatment of patients in overall less healthy populations (e.g. USA).

The Randomised Controlled Evidence

I’ve included a table below that you can skim and you’ll get the main messages without having to read too much messy detail. The overall impression is that despite there having been a previous negative trial from Holland in 2018, there are now three positive randomised trials that are changing the whole narrative.

Until recently, the only RCT was the 2018 Dutch study by Mahler et al. This small trial (55 patients) randomised patients with Kellgren-Lawrence grade 1–3 knee OA to 6 Gy versus sham radiotherapy using a LINAC.

There were some great things about this study. In particular, establishing a sham arm as the control was important. And they did it really well – they made it almost impossible for the patient to know which arm they were in, even going to the extent of playing audio of the sound of the LINAC to the sham radiotherapy group.

They reported outcomes using standard pain and function measures such as the VAS and OMERACT-OARSI composite score, with follow-up to 12 months. The results were negative – no meaningful difference between the treated and sham groups. The overall response rate in the two groups was 44% in the radiotherapy (LDRT) group and 43% in the sham group.

And their conclusion confidently stated:

“We found no substantial beneficial effect on symptoms and inflammatory signs of LDRT in patients knee OA, compared with sham treatment. Therefore, based on this RCT and the absence of other high-quality evidence, we advise against the use of LDRT as treatment for knee OA.”

Lets think about that a while before we even point out the issues with the trial. There are many thousands of patients in the literature, showing a consistent response rate of 70-90%. They then did a trial with 27 patients having radiotherapy and showing a poor response rate and then feel confident enough to state that we shouldn’t be treating these patients at all based on their trial. It seems like there’s something missing… Well, my suspicion is that it is politics and vested interests that pushed this conclusion, which could have been so much more balanced. For instance they could have said: “This small exploratory trial didn’t find an association, but we acknowledge the previous weight of data and recommend that larger randomised trials are performed to confirm this result”. And they could even have added “We welcome comments on why our trial conflicts historical large case series”. It feels like there was a certain arrogance that drowned out that sort of reasonable balanced view and any curiosity as to the incongruence.

Predictably there was an uproar from the experienced (mainly German) low-dose radiotherapy community that see daily responses in clinics round the country. I do find it slightly unfortunate that in their letter they used the metaphor of a tanker that dodges an iceberg, as we all know how vulnerable the Titanic was, despite its size and sophistication!

But they had serious points to make:

• Low numbers of patients, low follow-up of 3 months

• Many patients had long-standing pain (often for more than five years), which is likely to indicate a reduced chance of pain response

• Some had erosive OA

• The LDRT group had a higher BMI (perhaps indicating higher basal inflammation)

• The chosen dose (1 Gy × 6 fractions) may have been suboptimal compared to the now-favoured 0.5 Gy × 6 schedule.

• They did not offer a second phase of treatment if there was insufficient pain response after the first phase

And for me the most convincing aspect is that there was a poor response rate in both arms, way less than in historical case series.

Still, this trial, along with a sister trial by the same group (Minten et al) for hand OA became the cornerstone for scepticism about low-dose RT in OA.