Risk factors and the natural history of accelerated knee osteoarthritis: a narrative review

Knee osteoarthritis is typically a slowly progressive disorder. However, approximately 3.4% of adults develop radiographic evidence of accelerated knee osteoarthritis (AKOA) over 4 years [1, 2]. Therefore, at least 1 in 7 cases of incident knee osteoarthritis develop AKOA [1, 2]. We define AKOA as a process characterized by the rapid onset and progression from pre-radiographic disease to advanced-stage radiographic disease in less than 4 years (Kellgren-Lawrence [KL] grades = 0 or 1 to KL = 3 or 4) [1, 3, 4]. For the purpose of this review we will define the “onset” of AKOA as the first visit with radiographic evidence of advanced-stage radiographic disease. Individuals that develop AKOA typically progress from no or doubtful knee osteoarthritis (KL 0 to 1) to definite joint space narrowing and osteophyte (KL = 3) [5]. Two out of 3 adults that develop AKOA will experience this sudden onset and progression (KL 0 or 1 to KL 3 or 4) within 1 year [1, 3‐5].

Adults with AKOA represent an important proportion of adults with incident knee osteoarthritis. For example, at least 3 years before radiographic onset adults with incident AKOA have greater pain and disability compared to adults with a typical, gradual onset of knee osteoarthritis (KL 0 to 1, KL 0 to 2, or KL 1 to 2 over 4 years) [5, 6]. Furthermore, while very few people who develop typical knee osteoarthritis receive a knee replacement over 8 years (0.3%), more than 1 in 14 (7%) adults with AKOA undergo a knee arthroplasty within 2.3 years after initial signs of radiographic progression [7]. Including these adults in studies with those who develop typical knee osteoarthritis may yield misleading results in clinical trials and epidemiological studies [3]. Unfortunately, there are no comprehensive reviews to synthesize the risk factors and natural history for AKOA, as well as how AKOA compares with the current paradigm of typical knee osteoarthritis, aging (no radiographic knee osteoarthritis and no KL change over 4 years), and rapidly progressive forms of osteoarthritis. This latter point is particularly relevant because clinicians and researchers often interchange the terms accelerated and rapidly progressive osteoarthritis despite important differences between these disorders. The purpose of this narrative review is to summarize recent evidence from the Osteoarthritis Initiative about the risk factors and natural history of accelerated knee osteoarthritis (AKOA) – defined as a transition between no radiographic knee osteoarthritis to advanced-stage disease within 4 years – and put these new findings in context with typical osteoarthritis, aging, prior case reports/series, and relevant animal models. We acknowledge that the definition of typical knee osteoarthritis may be susceptible to misclassification because of a reliance on subtle changes in KL grades. However, we believe this definition has construct validity because the people with typical knee osteoarthritis differ from those without knee osteoarthritis based on reporting more knee symptoms the year before disease onset [6], reporting a knee injury more often before radiographic changes (a key risk factor for osteoarthritis) [1], and having more meniscal pathology [8].

To achieve our goal, we will start by reviewing the natural history of AKOA and its risk factors. The natural history will be divided into 3 phases: 1) 2 or more years before radiographic osteoarthritis onset, 2) 1 to 2 years before radiographic osteoarthritis onset, and 3) less than 1 year before radiographic onset. Within each phase, we will describe the clinical and structural changes that characterize each phase in comparison to adults with typical knee osteoarthritis or no knee osteoarthritis (representative of age-related changes). After summarizing the natural history of AKOA, we will describe the clinical consequences of AKOA, strategies to classify adults at risk for AKOA, and put AKOA into context with rapidly progressive osteoarthritis and destructive arthropathies.

Numerous factors may contribute to AKOA and help people identify who will develop it. Similar to typical knee osteoarthritis, older age and body mass index (BMI) are related to AKOA onset [1, 9, 10]. However, two subgroups are at greater risk for AKOA: 1) individuals < 65 years of age with BMI > 32.5 kg/m² and 2) individuals > 65 years of age that were typically overweight or obese with BMI < 35 kg/m² (only 27% had BMI < 25 kg/m²) [10‐12]. Greater age and BMI may contribute to the onset of AKOA through pathways related to hyperglycemia and elevated inflammation. However, contrary to this hypothesis, glycated serum protein concentrations, a biomarker of glucose homeostasis, associated with the subsequent onset of typical knee osteoarthritis but not AKOA [13]. There are also no statistical associations between serum concentrations of C-reactive protein or glucose and AKOA onset over the subsequent 4 years to those measurements [13]. It remains unknown if a change in biomarkers of systemic inflammation or glucose homeostasis occurs as people develop pre-radiographic structural changes that antedate radiographic disease onset.

Certain biomechanical factors may also contribute to AKOA. While the presence of static knee malalignment does not significantly differ between individuals who will develop AKOA or typical knee osteoarthritis, greater coronal tibial slope (describing the slope of the tibial plateau relative to a perpendicular line to the long axis of the tibia) is related with a greater odds of incident AKOA but not typical knee osteoarthritis when compared to adults without knee osteoarthritis. However, this relationship was only present among people who had a varus or valgus static malalignment [14]. Hence, a joint may tolerate aberrant coronal tibial slopes in isolation but become susceptible to failure when stressed by combining altered coronal tibial slope with static malalignment (i.e., varus or valgus malalignment).

At an individual level, we’ve described how adults with AKOA experience greater symptoms and functional impairments starting up to 3 years prior to radiographic disease onset. This complements findings that over the first 8 years of the Osteoarthritis Initiative, people who develop AKOA are more likely to report receiving arthroscopic knee surgery, intra-articular injections (i.e., hyaluronic acid, corticosteroids), and nonsteroidal anti-inflammatory drugs (over the counter or prescription) than people who develop typical or no knee osteoarthritis [45]. Even prior to the onset of radiographic evidence of AKOA, people who will develop AKOA are more likely to use certain prescription analgesics (e.g., acetaminophen, celecoxib, aspirin) and receive arthroscopic knee surgery than those who will develop typical knee osteoarthritis [45].

The greater personal burden of AKOA is further highlighted by the high rate of knee replacements. While most people who started the Osteoarthritis Initiative without radiographic knee osteoarthritis did not receive a knee replacement over the first 9 years of the study (≤ 1%), almost 1 in 7 knees with AKOA received a knee replacement during that period [7]. Furthermore, the median time from any increase in radiographic severity (change in KL grade) to knee replacement was 2.3 years (range 0.3 to 7.3 years) compared with the few knees with typical osteoarthritis that received a knee replacement (3.0 years; range 1.7 to 4.2 years) [7]. Adults with AKOA experience significant disease burden and have only a short window of opportunity to intervene, with 1 in 14 knees receiving a knee replacement in less than 2.3 years after the first evidence of radiographic progression.

Another concern is that our perception of osteoarthritis may be biased by our failure to account for people with AKOA. Within the Osteoarthritis Initiative, AKOA accounts for more than 1 in 5 cases of incident knee osteoarthritis [1]. Hence, factors that have a strong relationship with AKOA, but not typical knee osteoarthritis, may still appear associated with knee osteoarthritis overall. For example, excluding people with incident AKOA from analyses that defined incident knee osteoarthritis as at least a 2-point increase in radiographic severity led to smaller effect estimates (Cohen’s d [46]) when comparing cases of incident disease to controls: knee pain over time (d = 0.41 for everyone, d = 0.14 after excluding AKOA) or recent knee injury (OR = 5.4 for everyone, OR = 4.0 after excluding AKOA) [3]. It is essential that we critically assess our current conceptual model of knee osteoarthritis by accounting for differences among phenotypes or clinically relevant subsets (e.g., AKOA). Furthermore, we need to consider the consequences of recruiting people with AKOA into clinical trials. Separate trials may be necessary for AKOA and other types of knee osteoarthritis given AKOA has a distinct natural history defined by the onset of a destabilizing meniscal tear in a joint already susceptible to failure. A critical question is whether people with AKOA will respond differently to therapeutic interventions when compared to those with typical osteoarthritis.

To develop prevention strategies and clinical trials for adults at risk for AKOA, it is crucial to identify who will likely develop AKOA. Riddle and colleagues developed a prediction rule to estimate the chance of incident AKOA over the subsequent 5 years and found that the presence of possible knee osteoarthritis (KL = 1), contralateral knee osteoarthritis, greater BMI, and joint symptoms (total Western Ontario and McMaster Universities Osteoarthritis Index [WOMAC] score) increased the chance of someone developing AKOA [4]. While this prediction rule provided valuable insights into identifying people at risk for AKOA, it inadequately addressed the complex interactions among the risk factors, and it lacked MR-based structural features that may help classify individuals at risk for AKOA. Hence, we used classification and regression tree (CART) analyses, which can 1) identify the most statistically important factors and the associated cut points to most efficiently differentiate groups for classifying individuals, and 2) account for complex interactions that may optimize the ability to classify individuals at risk for AKOA. Furthermore, we used the CART analyses to explore if MR-based structural findings could help classify people at high risk for AKOA (54 people developed AKOA and 108 sex-matched Osteoarthritis Initiative participants). There is consistent evidence that age, fasting glucose concentrations, and static knee alignment were statistically important clinical factors to consider when classifying people at risk for incident AKOA [11, 12]. Effusion-synovitis volume and cruciate ligament degeneration may also be statistically important for classifying people at risk for AKOA over the subsequent 4 years [11]; however, adding data from magnetic resonance imaging (e.g., quantitative effusion-synovitis, cartilage damage, bone marrow lesions; semiquantitative assessments of menisci, tendons, ligaments) failed to improve the ability to classify adults who develop AKOA over the models with only clinical measures (Table 1). Furthermore, the models failed to explain the majority of the variance that contribute to someone being classified as incident AKOA or not. Hence, we are either missing a major factor that contributes to classifying people with AKOA or there are numerous factors that contribute a little to the classification. Going forward, it will be important to better understand the earliest phases of AKOA to develop strategies to identify people at risk for or with early-stage AKOA.

Numerous terms have been used during the last 60 years in the medical literature to describe a dramatic change in joint health; including rapidly progressive disease, rapidly destructive disease, accelerated degeneration, massive osteolysis, rapid chondrolysis, among others. The heterogeneity in nomenclature reflects uncertainty about how to classify rapidly destructive forms of osteoarthritis and their underlying mechanisms.

It is informative to reassess the existing literature and consider how various definitions may impact the reported epidemiology of rapidly progressive or accelerated osteoarthritis. For example, in Table 2, we offer several definitions of accelerated or rapidly progressive osteoarthritis and how the incidence using these definitions vary from 0 to 22% within the Osteoarthritis Initiative.

Recently, two classifications for rapidly progressive osteoarthritis have been adopted to characterize the rapid joint destruction observed in clinical trials for anti-nerve growth factor (NGF) agents [51, 52]. Rapid progressive osteoarthritis type 1 is similar to an early definition provided by Lequesne [50] that requires more than 2 mm of loss in joint space width in less than one year. Rapid progressive osteoarthritis type 2 is defined by abnormal bone loss or destruction (e.g., osteolysis) [51, 52]. Within the context of the anti-NGF clinical trials, these types of rapid progressive osteoarthritis usually (but not always) manifest in a joint in which osteoarthritis was already present [53]. In contrast, research focused on AKOA has focused on incident disease [4, 40, 54]. While these differences may be consequent on the differences in research design and setting, it remains unclear whether AKOA and rapid progressive osteoarthritis type 1 are separate or overlapping entities. A clear distinction between AKOA and rapid progressive osteoarthritis is that adults with AKOA rarely experience the rate of joint space loss nor the dramatic bone destruction observed with rapid progressive osteoarthritis. For example, no one with AKOA in the Osteoarthritis Initiative experienced more than 2 mm of loss in joint space width in less than one year nor abnormal bone loss or destruction. Furthermore, rapid progressive osteoarthritis is often described in knees, hips, or shoulders [46, 50, 53, 55, 56]. However, accelerated osteoarthritis is primarily observed in the knee and, to a lesser extent, the hand. Therefore, it is unclear how rapid progressive osteoarthritis at large ball-and-socket joints relates to findings at the knee.

Despite differences between AKOA and rapid progressive osteoarthritis, some interesting similarities may be worth exploring further. Both are more common among older adults and may be antedated by certain analgesic medications [45]. Furthermore, people who develop AKOA or rapid progressive osteoarthritis have early evidence of cartilage degradation followed by an extreme rate of articular cartilage loss [8, 31, 57] and early evidence of inflammation [11, 15, 16, 57]. Finally, 12% of knees that develop AKOA experience attrition or subchondral fractures [8] and 35% of people who develop accelerated hand osteoarthritis develop new central erosions in other hand joints [58]. While people with accelerated osteoarthritis never experience the dramatic bone destruction or collapse seen with some types of rapid progressive osteoarthritis it may be beneficial to further explore the role of bone changes in each type of osteoarthritis.

Until we reach greater clarity about the pathogeneses of these subsets of osteoarthritis, it may be inappropriate to conflate them in clinical practice or research studies. Furthermore, we need to be cautious about collapsing these phenomena together or extracting results from one type of osteoarthritis to another. Further studying the similarities and differences between AKOA and other forms of rapidly progressive osteoarthritis may yield new insights into these subsets of osteoarthritis.

Accelerated osteoarthritis is distinct from typical osteoarthritis. The incidence of AKOA is likely greater than commonly perceived and could have a profound impact on epidemiologic studies and clinical trials. There is an urgent need to consider this subset of osteoarthritis when performing clinical research and to create standard nomenclature for the array of arthropathies that may be related but distinct from AKOA.