Over the past 5 years, we have witnessed significant advances in the field of osteoarthritis (OA) in terms of epidemiology, prognostic markers and therapeutic options. The year 2018 can be considered a breakthrough with the publication of a draft guidance by the Food and Drug Administration (FDA) designating OA as a serious disease.1 This is a revolution in the field, as this designation opens up the possibility for pharmaceutical companies to apply for accelerated approval during the development of a disease-modifying OA drug (DMOAD) based on yet-to-be defined surrogates. The Annals of the Rheumatic Diseases (ARD) has been instrumental in disseminating clinical research on OA, the world’s most common chronic joint disease. To further highlight the most recent developments in the field, this collection brings together some of the most outstanding original articles (reviews, editorials, viewpoints and letters have been excluded) published in ARD over the last 5 years dealing with clinical research in OA (animal studies and pathophysiology-focused articles have been excluded).

Analysis of data from the Global Burden of Diseases, a study that provides a comprehensive picture of mortality and disability in >160 countries for >350 health conditions, shows that between 1990 and 2017, the incidence of OA increased by 8.2% (95% uncertainty intervals (UI) 7.1% to 9.4%) from 1990, at an annual rate of 0.32% (0.28% to 0.36%), reaching 181.2 (95% UI 162.6 to 202.4) per 100 000 inhabitants.2 3 It was particularly pronounced in regions with a low sociodemographic index. The prevalence of OA was 3754.2 per 100 000 (95% UI 3389.4 to 4187.6), an increase of 9.3% (95% UI 8% to 10.7%) from 1990. There were about 303.1 million (95% UI 273.3 to 338.6 million) prevalent cases of hip and knee OA. In 2017, 9.6 million years lived with disability were recorded, making it one of the most disabling diseases in the world. Physical disability leads to walking difficulties when it affects the knees or hips, thus increasing cardiovascular events due to a lack of physical activity. The study of three cohorts by Zeng et al is the first to show such a clear and significant association between knee OA and deep vein thrombosis, with a multivariable-adjusted HR of 1.38 (95% CI 1.23 to 1.56).4

The reasons why the incidence of OA is increasing worldwide seem clear, with the obesity epidemic at the forefront, in addition to the ageing of the world’s population. It is interesting to note that this risk is aggravated by the inadequate adaptation of our organism to the rapid environmental changes in our lifestyles, which places OA in the group of the mismatch metabolic disorders. Further evidence of this was provided by the highly original study by Wallace et al, who observed an increased susceptibility to knee OA in the indigenous, non-industrial Mexican Tarahumara population due to rapid lifestyle changes.5 Despite significantly lower levels of obesity than Americans, the Tarahumara accumulated greater abdominal adiposity for a given body weight, which was associated with an increased risk of radiographic and symptomatic knee OA for a lower body mass index than Americans.

The FDA’s accelerated approval process for a future DMOAD is based on the ability to rely on one or more surrogate markers that can predict its effect on a hard endpoint, which could be the delay to total knee replacement (TKR). TKR is essentially related to symptom severity, although structural damage is also part of the decision. The critical question is whether or not structural joint changes correlate with symptoms. Using 600 subjects from the Osteoarthritis Initiative (OAI), a cohort studying >4700 knee patients with or at-risk for OA with data on imaging, clinical and participant factors, Bacon et al quantified the association between cartilage loss and worsening knee pain after adjusting for bone marrow lesions (BMLs) and synovitis, and examined the extent to which these factors mediated this association.6 They observed a significant but very weak association between reduced cartilage thickness and worsening pain over 24 months, largely mediated by the degree of synovitis (but not BMLs). Their provocative conclusion is: ‘Demonstrating that chondroprotection reduces knee pain will be extremely challenging and may be unattainable’. As a result, other groups are searching for new imaging biomarkers that could better predict worsening symptoms and TKR. This is the case, for example, with the study of Bowes et al, which measures three-dimensional bone shape based on AI analysis of MRI images, resulting in 40 different scores compared with the classic radiological 4-score Kellgren Lawrence (KL1–4).7 For the same radiological KL3 score, the authors found six different bone shape scores, some of which were significantly more associated with pain exacerbation than others.

The use of liquid biomarkers as surrogates in a therapeutic DMOAD efficacy trial remains very challenging because of the complexity of finding one or a combination of several of biomarkers whose change parallels OA improvement. Nevertheless, we see very active research in this area, such as the study conducted by the Foundation for the National Institutes of Health-OAI consortium.8 Thanks to this consortium, we already have promising options for predicting pain and structural deterioration. Innovative approaches are also being tested. Dvir-Ginzberg’s group described the serum N Terminal/C Terminal sirtuin1 (NT/CT SIRT1) ratio, which reflects both early OA and chondrosenescence.9

The search for new treatments has taken centre stage over the past 5 years due to the significant unmet needs of patients with OA who rarely find a solution to their symptoms with no definitive disease-modifying drugs on the market. This explains why many therapeutic ‘solutions’ are still pursued in the absence of strong scientific evidence, as is the case with low-dose radiotherapy in many countries, which will certainly be curtailed by this negative randomised sham trial published by Mahler et al.10

All current recommendations emphasise the essential role of non-pharmacological treatments, and in particular the importance of education and physical exercise. A randomised clinical equivalence trial in patients with knee OA compared over 8 weeks 102 patients who received an educational programme (2×1.5 hour classes) combined with physical exercise (12×1 hour classes) with 104 patients who received four intra-articular injections of physiological serum.11 In both groups, there was an improvement in the Knee Injury and Osteoarthritis Outcome Score for pain and function, with no statistical difference between the groups (2.7 (95% CI −0.6 to 6.0)), leading the authors to conclude that the interventions were equally effective. This highly original study raises more questions than it answers, such as the challenge of conducting this type of study in a blinded fashion, the role of the contextual effect of infiltrative procedures, the regression to the mean effect, challenges with compliance with education and exercise programmes, etc.

The search for new molecules capable of alleviating symptoms or even delaying the progression of OA remains a Holy Grail. As far as symptomatic treatments are concerned, much hope was pinned on tanezumab, an anti-nerve growth factor (NGF) antibody. For the first time, it seemed possible to offer our patients with knee/hip OA a treatment that was both easy to use (one subcutaneous injection every 8 weeks) and highly effective, seemingly surpassing the efficacy of non-steroidal anti-inflammatory drugs. Three pivotal phase III randomised, placebo-controlled clinical trials were conducted, two in the USA and one in Europe and Japan, the latter being published in the ARD.12 Unfortunately, a yet unexplained safety issue—a risk of rapid progressive osteoarthritis of approximately 1.4-2.8 per 100 patient-years at a dose of 2.5 to 5 mg every 8 weeks—led to a negative opinion by the FDA and European Medicines Agency based on what they considered an unfavourable benefit-risk balance. This unexpected safety issue underscores the importance of pharmacovigilance, which is not without its surprises. Two examples in the area of OA include the finding in a UK registry of a 50% increase in the risk of total knee or hip replacement in people taking warfarin, a widely prescribed anticoagulant.13 Even more surprisingly, a study in the famous UK Biobank database showed a 15% reduction in all-cause mortality with long-term glucosamine use.14 Of course, all these observational studies must be interpreted with caution, as there may be several confounding factors that have not been taken into account.

Another innovative biotherapy that failed in the recent years was lutikizumab, an anti-interleukin (IL)-1 with the property of targeting both IL-1α and IL-1β. This molecule showed no efficacy in erosive hand OA (EHOA) despite subcutaneous injection every 2 weeks in a 24-week randomised placebo-controlled trial.15

Regarding DMOADs, there has been great interest in a cartilage growth factor, fibroblast growth factor 18 or sprifermin, which, when injected intra-articularly, raised hopes for delaying the progression of OA, but failed to reach the symptomatic endpoint after 2 years of treatment, even at the highest dose every 6 months. Nevertheless, it is remarkable to see for the first time a molecule capable of increasing cartilage thickness in some areas and reducing cartilage degradation in others by MRI analysis up to 5 years after the first injections, making it the longest clinical trial ever conducted with a molecule seeking to become a DMOAD.16 17 However, the story is not over yet, as it cannot be ruled out that the symptomatic effect may still be seen in certain subgroups of patients, paving the way for further clinical trials.

Will the solution come from drugs already on the market for other indications? In other words, can we reposition certain drugs to treat OA? Undoubtedly, we have been disappointed in recent years by the failure of many biotherapies that are known to be effective in rheumatoid arthritis but unable to demonstrate their efficacy in OA. This was again the case with tocilizumab, which targets IL-6R, in a negative clinical trial in erosive hand OA, despite EHOA being considered the most inflammatory phenotype of OA.18 However, other avenues are being explored, such as glucagon-like protein receptor agonists (GLP1RAs) administered subcutaneously in patients with diabetes or for weight management. Zhu et al analysed data on patients with diabetes from a very large cohort of >40 000 Chinese patients with knee OA. They compared the level of pain, the velocity of cartilage degradation as measured by MRI and the number of knee surgeries among patients with diabetes with knee OA treated or not with GLP1RAs. They observed a significant association between GLP1RA use and all endpoints.19 Interestingly, because of the other property of these molecules to facilitate weight loss, they looked at this causal role in mediating this association and found that only 31% of this association was mediated by weight loss suggesting an additional direct effect on joints. Another exciting study, based on a very large UK database, suggests that digoxin may delay prosthetic surgery, giving it disease-modifying properties.20 This drug emerged from their screening analysis of >1000 molecules already approved by the FDA, selected on the basis of their ability to stimulate chondrocyte anabolism.

Although not yet reflected in current practice, considerable progress has been made in clinical research in OA. The extent to which the ARD has kept pace with this trend is evident from the quality of the original studies published over the past 5 years. There is no doubt that the collection for the next 5 years will be just as rich. And we cannot wait to discover it!

Not applicable.

Not applicable.