Although X-ray images are still a mainstay of imaging in the radiology department, in pharmaceutical trials for rheumatoid arthritis (RA) and osteoarthritis (OA) therapies, their use is becoming increasingly limited. In fact, they are a bit like that old fax machine in the corner – still useful and sometimes necessary but becoming less relevant. So, what is taking the place of X-rays and why? In this article, we will explore the increasing use of MR images in these trials, how these images can be automatically analysed using machine learning algorithms; and the reasons that MRI is displacing X-rays.
The major issue facing the use of X-ray images in the demonstration of structural change is that most subjects with RA will already be on some form of therapy before they get enrolled into a trial. This means that the amount of structural damage to the joints seen in a baseline hand X-rays will be relatively minor, and because the damage to joints due to RA is irreversible, trials are conducted an active comparator rather than a true placebo, meaning that the amount of structural change seen in follow-up X-rays is very small. If we look at OA trials, the problem of using X-rays to measure joint space narrowing (JSN) in the knee is one of attempting to measure a small change that happens slowly. Given that JSN due to OA is around 0.2mm a year, and the measurement error of JSW is around 0.6mm, it is very important to control the position of the knee so that the same projected image is produced each time.
The advantage that MRI has over X-ray imaging is two-fold: it produces images of soft tissues and soft tissue changes in addition to bony changes; and it provides truly 3D image data which allows the anatomy to be fully understood. Machine vision algorithms can be used to automatically analyse these MR images to produce quantitative results. So, what does it mean for the clinical trials landscape?
- Smaller Trials
The use of MRI to image soft-tissue changes such as synovitis and bone marrow oedema in RA trials; and cartilage thickness or bone area changes in OA trials means that researchers can see the effects of a new drug faster and with smaller cohorts of patients. This can reduce commercial risk of a trial because patient recruitment is one of the leading cost drivers of these trials, and sponsors and pharma companies have earlier information to inform critical go/no-go decisions. Quantification of these imaging changes by computer algorithms improves things even further, with reproducibility that is superior to the intra-observer variability seen between expert radiologists.
- Better Quality Data
MRI produces truly 3D image data which provides accurate insight into the anatomical changes due to disease progression. By contrast, X-rays are, by their nature limited to 2D and often, anatomical changes can be hidden. It is notable, that in studies comparing MRI and X-ray imaging, many more RA erosions are seen using MR than are detected by X-ray. In the case of OA, trails, osteophytes on the edge of the joint seen in an X-ray are often used as important inclusion criteria r subjects. MR images reveal that osteophytes are in fact a ridge of hard tissue surrounding the joint and prominent ridges on the anterior or posterior of the joint can be missed in an X-ray since these will be hidden.
- Novel Measurements
This is probably where the use of MRI makes the most powerful contribution to the RA and OA clinical trials landscape. X-ray is well known in its limitations to measure disease progression and changes. In addition, there are anatomical changes, such as shape and area change, that radiologists can’t measure manually and therefore, accurately. Using MRI, Imorphics technology is able to identify and measure bone shape changes as a novel biomarker in clinical trials. This is a hugely important for OA trials, giving researchers a new, reliable and automated way to understand the disease and introduce innovative medicines and treatment options in future.
As MRI becomes more important in clinical trials, the number of images to process grows and getting these images read using traditional methods of manual review by a radiologist drastically limits the ability to be agile in the current (and future) environment. But applying automation with excellent reproducibility, research teams can turn images into actionable information quickly – meaning shorter time scales, fewer patients and reduced costs. Imorphics has demonstrated the impact of automation in comparison to expert radiology reading in RA trials where researchers could see the effect of drugs in around one month with a small cohort of patients per treatment arm.
- Population health
We hear a lot about big data and its impact on population health. While there’s a huge amount of work being put into understanding how to apply this to various aspects of population health, there’s an opportunity in OA trials to turn big image data analysis problems into actionable information. I’ll be talking about this in more detail in my next post, follow our LinkedIn page to get updates or contact me on firstname.lastname@example.org to talk about how Imorphics can redesign your next clinical trial for the better.
The fax machine still has its place, but email is so much better.