There are many different types of arthritis, and diagnosing the exact type of inflammatory disease that is affecting a patient’s joints is not always easy. In an interdisciplinary research project conducted at Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) and Universitätsklinikum Erlangen, computer scientists and physicians have now succeeded in teaching artificial neural networks to differentiate between rheumatoid arthritis, psoriatic arthritis and healthy joints.

Within the scope of the project funded by the Federal Ministry of Education and Research (BMBF) called “Molecular characterization of arthritis remission (MASCARA)”, a team led by Prof. Andreas Maier and Lukas Folle from the Chair of Computer Science 5 (Pattern Recognition) and PD Dr. Arnd Kleyer and Prof. Dr. Georg Schett from Department of Medicine 3 at Universitätsklinikum Erlangen had the remit to investigate the following questions: Can artificial intelligence (AI) detect various types of arthritis using joint shape patterns? Does this method allow us to make more precise diagnoses in cases of undifferentiated arthritis? Are there certain areas in joints that should be examined in more detail during a diagnosis?

Missing biomarkers currently often make precise classification of the relevant type of arthritis difficult. X-ray images used to aid diagnosis are not completely reliable either, as their two-dimensionality is not precise enough and leaves room for interpretation. This is in addition to the fact that positioning the joint being examined for an X-ray image can be difficult.

Artificial networks learn using finger joints

To find the answers to its questions, the research team focused its investigations on the metacarpophalangeal joints of the fingers – regions in the body that are very often affected early on in patients with autoimmune diseases such as rheumatoid arthritis or psoriatic arthritis. A network of artificial neurons was trained using finger scans from high-resolution peripheral quantitative computer tomography (HR-pQCT) with the aim of differentiating between “healthy” joints and those from patients with rheumatoid or psoriatic arthritis.

HR-pQCT was selected as it is currently the best quantitative method of producing three-dimensional images of human bones in the highest resolution. In the case of arthritis, changes in the structure of bones can be very accurately detected, which makes precise classification possible.

Neural networks could make more targeted treatment possible

A total of 932 new HR-pQCT scans from 611 patients were then used to check if the artificial network can actually implement what it had learned: Can it provide a correct assessment of the previously classified finger joints?

The results showed that AI detected 82% of the healthy joints, 75% of the cases of rheumatoid arthritis and 68% of the cases of psoriatic arthritis, which is a very high hit probability without any further information. When combined with the expertise of a rheumatologist, it could lead to much more accurate diagnoses. In addition, when presented with cases of undifferentiated arthritis, the network was able to classify them correctly.

“We are very satisfied with the results of the study as they show that artificial intelligence can help us to classify arthritis more easily, which could lead to quicker and more targeted treatment for patients. However, we are aware of the fact that there are other categories that need to be fed into the network. We are also planning to transfer the AI method to other imaging methods such as ultrasound or MRI, which are more readily available,” explains Lukas Folle.

Hotspots could lead to faster diagnoses

Whereas the research team was able to use high-resolution computer tomography, this type of imaging is only rarely available to physicians under normal circumstances because of restraints in terms of space and costs. However, these new findings are still useful as the neural network detected certain areas of the joints that provide the most information about a specific type of arthritis that are known as intra-articular hotspots. “In future, this could mean that physicians could use these areas as another piece in the diagnostic puzzle to confirm suspected cases,” explains Dr. Kleyer. This would save time and effort during the diagnosis and is already in fact possible using ultrasound, for example. Kleyer and Maier are planning to investigate this approach further in another project with their research groups.

Dental Health Tips for Rheumatoid Arthritis Patients

The incidence of dementia in patients with rheumatoid arthritis is lower in patients receiving biologic or targeted synthetic disease modifying antirheumatic drugs (DMARDs) than in patients who receive conventional synthetic DMARDs, according to a new study. The study was presented at the virtual annual meeting of the American College of Rheumatology.

“Being on a biologic or targeted synthetic DMARD actually decreased your risk of incidence of dementia by 17% compared to patients who were on a conventional synthetic DMARD only,” said lead study author Sebastian Sattui, MD, MS, a rheumatology fellow at Hospital for Special Surgery (HSS) in New York City. The study was done in collaboration with investigators from Weill Cornell Medicine and the University of Alabama at Birmingham.

Dr. Sattui said that the treatment in patients with rheumatoid arthritis has become more complex based on the understanding that rheumatoid arthritis has an impact well beyond what are thought of as the classical manifestations. Previous studies have suggested that inflammatory diseases such as rheumatoid arthritis can increase the risk for dementia and that TNF agents may have a role in preventing the incidence of dementia.

In the new study, researchers identified a cohort of patients with rheumatoid arthritis in Medicare claims data from 2006 to 2017. To be eligible, patients had to have continuous enrollment of at least 12 months in Medicare Part A, B and D, be at least 40 years of age and have no prior diagnosis of dementia.

In the sample of 141,326 eligible patients with rheumatoid arthritis, the crude incident rate of dementia was 2.0 per 100 person-years for patients on conventional synthetic DMARDs and 1.3 for patients on any biological DMARD. After adjusting for factors such as age, sex and other comorbidities, patients on biologic or targeted synthetic DMARDs had an adjusted 17% lower risk for dementia than patients on conventional synthetic DMARDs. No significant differences were observed between the different classes of biologic or targeted synthetic DMARDs, suggesting that decreased risk is possibly explained by the overall decrease in inflammation rather than a specific mechanism of action.

The researchers say clinicians should factor this new information into treatment decisions, but prospective studies are needed. “Our work shows yet another dimension in which treatment of rheumatoid arthritis can impact the overall health and quality of life of our patients,” said Dr. Sattui. “Rheumatoid arthritis is a systemic disease and it can have cognitive implications. However, these complications seem to share similar pathways to those of articular disease, and the medications that we use to treat rheumatoid arthritis could be effective in the prevention of dementia in patients with rheumatoid arthritis. Future studies need to assess the impact of the interventions, such as the treat-to-target strategy, on the incidence of dementia in patients with rheumatoid arthritis.”

Scientists at deCODE genetics, a subsidiary of Amgen, have together with their Nordic collaborators published the largest genome-wide association study to date on rheumatoid arthritis (RA) in Annals of the Rheumatic Diseases, including over thirty thousand cases and one million controls.

RA is the most common inflammatory joint disease and most patients need lifelong immunosuppressive therapy. The study is based on an extensive collaboration between researchers in six northwestern European countries and includes information on not only RA overall but also the disease subsets, defined by serology (rheumatoid factor/anti-CCP antibodies). A total of 64 million sequence variants were investigated, based on whole-genome sequencing of a large number of individuals from these countries. It was determined thereafter whether sequence variants that associate with RA or its subsets affect protein coding, gene expression and/or levels of five thousand other proteins in plasma.

Several previously unreported sequence variants were found to have large effect on the risk of seropositive RA, while associations with seronegative RA were scarce. Through sequential application of genomics, transcriptomics and proteomics, causal genes were identified for most signals and the majority of those that associate with seropositive RA encode proteins in the network of interferon-alpha/beta and IL-12/23 that signal through the JAK/STAT-pathway. This includes a missense variant in the STAT4 gene that confers 2.27-fold risk, larger than previously reported signals, and it leads to a replacement of hydrophilic glutamic acid with hydrophobic valine in a conserved, surface-exposed loop of the STAT4 protein. Furthermore, a stop-mutation in the FLT3 gene increases seropositive RA risk 35%, while three missense variants in the TYK2 gene confer 15-59% reduced risk and affect levels of the interferon-alpha/beta receptor 1 (IFNAR1).

These findings highlight how a multiomics approach can reveal causal genes. Furthermore, they support treatment of seropositive RA with the already registered JAK and IL-6R inhibitors as well as CTLA4-Ig, but also open for repurposing of other drugs that target proteins in the JAK/STAT-pathway, including inhibitors of FLT3, TYK2 and IFNAR1, that are currently used or under development for other diseases.