Abstract
Osteoarthritis (OA) remains a major epidemiological problem worldwide. The lack of established chondroprotective agents limits the clinical management of this disease to relief of symptoms, such as pain and joint stiffness. The marked increase in the prevalence and incidence of OA among women after the menopause draws attention to the possible contribution of estrogen deficiency to acceleration of cartilage degradation. Hormone replacement therapy (HRT) seems to counter the adverse influence of ovariectomy on the progression of joint lesions in various animal models, including nonhuman primates. Although decreases in the risk of OA associated with long-term HRT were repeatedly observed in large-scale epidemiological studies, the chondroprotective potential of this therapeutic option has received modest recognition among healthcare professionals. The overall diversity of results hampering our ability to draw conclusions at the present time arises from the diversity of methodological approaches and selected outcome measures of OA. Randomized clinical trials using state-of-the-art methodology, such as high-resolution magnetic resonance imaging that allows direct visualization and morphometric characterization of articular cartilage and biochemical markers of cartilage turnover that can provide early indication of therapeutic effects, are awaited to obtain evidence-based answers to the utility of estrogen in the prevention of OA.
As defined by Altman and colleagues: “Osteoarthritis is a heterogeneous group of conditions that lead to joint symptoms and signs which are associated with defective integrity of articular cartilage in addition to related changes in the underlying bone at the joint margins” [1].
The hallmark of osteoarthritis (OA) is progressive cartilage and joint degradation. In most cases, OA is diagnosed at a late stage when the progressive disease has already caused major radiographic alterations. This might be due to a relatively symptom-free course of cartilage degradation. The painful and disabling symptoms are more frequently associated with large subchondral edema lesions, synovial hypertrophy and large synovial effusions, as recently demonstrated by magnetic resonance imaging (MRI)-assisted investigations [2].
OA is the most common joint disorder in the world. In the USA, it is second only to ischemic heart disease as cause of work disability in people over 50 years of age [3]. There are notable gender differences in the prevalence and severity of OA. Particularly after the menopause, women are more likely than men to develop generalized OA and, interestingly, women have a tendency to develop a painful and deforming form of OA in the hand, known as ‘menopausal arthritis’ [4].
The prevalence of radiographic OA rises steeply with age at all joint sites. In a large population-based observational study, which included 6585 participants, a total of 75% of women aged 60–70 years had OA in their distal interphalangeal (DIP) joints, and even by 40 years of age 10–20% of subjects had evidence of severe radiographic disease of their hands or feet [5]. Knee OA appears less frequent than OA of the hand and foot. Both hand and knee disease appear to be more frequent among women, although the female-to-male ratio varies across studies from 1.4–4.0 [6,7]. Hip OA is less common than knee OA, and prevalence rates for the two genders seem similar.
The most recent data approximating the incidence of symptomatic hand, knee, and hip OA were obtained from the Fallon Community Health Plan in the USA [8]. In this study, the age-and sex-standardized incidence rate of hand OA was 100/100,000 person-years; for hip OA 88/100,000 person-years; and for knee OA 240/100,000 person-years. The incidence of OA is higher in men than women before 50 years of age [9] but later, women have higher incidence rates than men, indicating a likely contribution of estrogen deficiency [8].
As longevity increases, so will the prevalence of OA. In the USA alone, there are approximately 43 million individuals (one in six) with OA, and most are older than 45 years. By the year 2020, 59.4 million people in the USA will be affected by OA, thus increasing chronic disability and costs by more than 25%. The annual cost to society in medical care and lost wages can now be estimated to exceed US$100 billion [10]. Therefore, physicians who provide care for the increased number of patients with arthritis must be aware of improved therapeutic modalities to reduce arthritis-related disabilities and hospitalizations, and to preserve function.
Estrogen deficiency & osteoarthritis
The marked female predominance of polyarticular OA, and in particular the marked increase of OA in women after the menopause supports the notion that female sex steroids play an important role in the maintenance of cartilage homeostasis. This theory generated a new field of research investigating the effects of estrogens on chondrocytes and joint diseases. The ultimate practical aim is to clarify whether estrogen replacement therapy (ERT) could provide benefits in preventing or even halting the progression of OA in postmenopausal women. The current article attempts to summarize the preclinical and clinical evidence supporting the potential of this therapeutic option.
Results from preclinical studies
The role of estrogen in vivo can be addressed by investigating the impact of eliminating endogenous resources or the effects of treatment with exogenous estrogen. The authors' research group recently reported that surgical removal of the ovaries evoked pronounced increases in the urinary excretion of the C-terminal telopeptide of collagen Type II (CTX-II), a biomarker of cartilage turnover, compared with sham-operated animals [11]. This acceleration of cartilage degradation culminated in a more rapid development of erosions in the articular cartilage of the knee. The histological score of structural damage was proportional to the magnitude of the early increase in the biomarker. Similarly accelerated progression of joint destruction in ovariectomized rats was also documented by independent investigators [12].
In this model of ovariectomized rats, both estrogen and levormeloxifen (a selective estrogen receptor modulator [SERM]) were able to restore cartilage turnover to levels present in sham-operated animals and reduce the propagation of articular cartilage damage [13]. When assessed in larger animals, such as sheep, ovariectomy evoked detrimental effects on the intrinsic material properties of the articular cartilage of the knee, even though there was no apparent difference in cartilage thickness compared with sham-operated animals 12 months after ovariectomy [14].
However, the administration route and/or the dose of estrogen to which chondrocytes were exposed appears to influence the therapeutic benefits. Intra-articular treatment with higher doses of estradiol for 12 weeks was associated with a damaging effect (loss of condyle surface congruity, thinning, fissuring, and fibrillation of the articular cartilage) and later, changes in the subchondral bone tissue were observed. The same dosing regimen using low-dose estradiol did not seem to provoke this structural harm [15]. The same group advocated intra-articular dosing of SERMs as a treatment option at the early stage of OA [16].
Extensive research has characterized the long-term effects of estrogen in nonhuman primates, which offer a particularly useful model to approximate the disease in humans. Cynomolgus macaques spontaneously develop OA, and the histological lesions in the tibial plateaus and femoral condyles are very similar to those in humans [17]. In female cynomolgus macaques that underwent bilateral ovariectomy, 3-year treatment with conjugated equine estrogen significantly reduced the severity of OA compared with nontreated animals. The treatment effects at the end of the treatment period were analyzed at 4 levels. Level 1 described differences in OA severity (articular cartilage structure, number of chondrocyte clones, loss of toloudine blue staining), level 2 described differences in subchondral bone thickness, level 3 described the variability in articular cartilage thickness and area, and level 4 described calcified cartilage areas and thickness. The investigators revealed significantly lower level 1 and higher level 2 scores (although the number of osteophytes was significantly lower) in estrogen-treated animals compared with controls. Assessment at level 3 and 4 did not indicate significant treatment-related benefits. Taking into account the large number of monkeys included in the study, the long duration of estrogen replacement, and the advanced and thorough methodology, this study raises a noteworthy argument for the protective effects of HRT against OA [18].
An important observation, recently reported by the same group, is that bilateral ovariectomy evokes pronounced increases in dynamic histomorphometric indices of bone turnover in the subchondral, but not the epyphyseal/metaphyseal cancellous bone compartment [19]. The investigators suggested intimate links between the acceleration of subchondral turnover and cartilage homeostasis, with direct implications for the pathogenesis of OA. Thus, part of the favorable impact of estrogen might be mediated by the well-established antiresorptive effect of estrogen. This notion is also supported by the observation that other antiresorptive agents, such as bisphosphonates, were also proposed as OA-modifying agents [20], yet estrogens can not only act as antiresorptive agents but can also exert direct favorable effects on chondrocytes [21].
Results from clinical studies
Endogenous estrogen & osteoarthritis
Numerous small and larger studies have correlated sex hormone levels to self-reported or radiologically verified OA. Samanta and colleagues investigated the relationship between endogenous sex steroid levels and self-reported OA, finding no support for association [22]. Similar conclusions were reached by Cauley and colleagues, who related endogenous estradiol levels to radiographic presence and severity of OA in 229 women with a mean age of 74 years [23]. Spector and colleagues reported increased levels of bioavailable estradiol and androgens in those with OA, suggesting harmful effects of excessive sex steroids exposure [24]. In a small study including 21 patients, Tsai and colleagues found an apparent strong association of excessive synovial estradiol and estrogen receptor binding with OA of the knee [25].
The aforementioned associations, however, can be confounded by numerous pathogenic factors. For example, we have recently shown that the highest levels of bioavailable estradiol can be measured in elderly women with upper-body obesity, a phenotype that is prone to insulin resistance, dyslipidemia, low-grade inflammation, diabetes, and accelerated atherogenesis [26]. Thus, we cannot exclude the possibility that the apparent lack of beneficial effects of endogenous estradiol on cartilage homeostasis actually reflect the presence of pathogenic factors with strong and competing adverse effects on articular cartilage.
HRT & osteoarthritis
Table 1 summarizes major studies investigating the interaction between HRT and OA as defined by either self-reported (questionnaires/symptoms) or radiographic measures.
Diversity of conclusions in clinical studies assessing associations between hormone replacement therapy and OA as defined by symptoms or radiography.
When using a symptom-based endpoint, the conclusions are diverse, yet when estimating structural effects with imaging techniques, most studies conclude benefits.
MRI: Magnetic resonance imaging; OA: Osteoarthritis.
Studies with self-reported symptoms as end-points
Von Muhlen and colleagues investigated associations between postmenopausal estrogen use and the prevalence of OA in a community-dwelling sample of 1001 postmenopausal women aged 43–97 years (mean 72 years) [27]. The diagnosis of OA was based on pain history and a clinical examination assessed by a trained study nurse. Postmenopausal estrogen use lasting for at least 1 year (average duration 14.1 ± 10.1 years) was associated with small but statistically significant increases in the prevalence of hip (4.1 vs 1.1%) and hand (15.8 vs 13.5%) OA, but not knee OA. The study by Maheu and colleagues investigated and compared various measures of disease activity between HRT users and HRT nonusers in 711 postmenopausal women 50–75 years old; 238 women had ‘painful’ hand OA, 240 ‘quiescent’ OA, and 233 served as controls [28]. They found no apparent influence of HRT on the severity or symptoms of hand OA. Similarly, using knee pain or the Western Ontario and McMaster Universities osteoarthritis index (WOMAC) score as an outcome measure, 4-year HRT did not seem to exert beneficial effects in a subgroup analysis of the Heart and Estrogen/progestin Replacement Study (HERS), which included elderly women with prevalent cardiac disease [29]. Collectively, these findings suggest that HRT has limited beneficial effects in terms of improving joint symptoms when administered to the elderly, in whom cartilage degradation due to aging and many years of estrogen deficiency has advanced considerably. However, as pointed out by investigators of the latter study, these conclusions may not apply to other groups of women or to the effect of HRT on the structural integrity of articular cartilage.
Findings of the Italia study seem to support this concept [30]. The study included data from a total of 42,464 women with a mean age of 53 years. Using a standard questionnaire, women were asked whether they suffer from OA, which was diagnosed by a physician, or requires medical or surgical treatment. A total of 12,521 women reported OA. Considering the implications of estrogen deficiency, women in their natural or surgical menopause were at slightly increased risk of OA with respective odds ratios (OR)s of 1.13 (95% confidence interval [CI] 1.07–1.21) and 1.18 (95% CI 1.08–1.28). Ever-users of HRT were at decreased risk of OA compared with never-users (OR 0.73, 95% CI 0.69–0.78). These observations thus support the hypothesis that HRT initiated in the early menopausal years might offer a therapeutic window for utilizing effective prevention of OA.
Studies with radiographic changes as end-points
Cross-sectional studies/analyses
Dennison and colleagues addressed associations between the presence of OA and use of estrogen in a population-based case–control study including 413 female cases (i.e., women >45 years old listed for hip replacement) and 413 age- and sex-matched controls [31]. The risk of hip OA was significantly elevated among women who had undergone bilateral oophorectomy (OR = 1.9, 95% CI 1.0–3.7). After adjustment for independent risk factors of hip OA and for other reproductive variables, long-term HRT use (>5 years) was associated with a 40% reduction in risk (OR = 0.6, 95% CI 0.2–1.8). Paradoxically, short-term HRT use was associated with an apparent increased risk for hip OA (OR = 1.7, 95% CI 0.9–3.3). Postmenopausal HRT use was relatively low in this cohort (short-term: 7.1%, long-term: 2.5%), which is also reflected by the broad CIs, limiting the strength of these conclusions. Nevertheless, the increase in risk for OA associated with oophorectomy supported the authors' conclusion that long-term HRT likely provides protection.
In the Ulm Osteoarthritis Study, with a somewhat similar study design, Erb and colleagues assessed associations between HRT and radiographically defined patterns of OA [32]. Participants were 475 women with a mean age of 66.1 years who underwent hip or knee joint replacement because of advanced OA. Patients were categorized as having bilateral or unilateral OA according to the presence or absence of radiographic OA in the contralateral joint. If radiographic OA of different hand and finger joint groups was present, participants were categorized as having generalized OA. Fifty–five women (11.6%) were using HRT for a median duration of 5.4 years. After adjustment for potential confounding factors, there were no signs of protective effects associated with HRT use. ORs for bilateral OA and generalized OA were 1.21 (95% CI 0.48–3.03) and 1.21 (95% CI 0.53–2.74), respectively. However, several limitations weakened the analysis. First, information was obtained for current regular medication during the past 3 months only. Thus, no information on past use of HRT was collected, which could potentially lead to the inclusion of some past users in the group of nonusers. Furthermore, no information was collected that could have allowed the investigation of the time sequence between HRT and OA patterns. Therefore, the possibility of an OA manifestation before the initiation of HRT use, or the notion that HRT was used by women to relieve symptoms of painful joint disease could not be ruled out.
In the Study of Osteoporotic Fractures, Nevitt and colleagues investigated the association between HRT and radiographic findings of OA of the hip [33]. This is currently the largest cross-sectional study, including validated data on 4366 postmenopausal women aged over 65 years old. Radiographs of the pelvis were assessed for structural features of OA of the hip on a summary scale of 0 (none) to 4 (severe OA). Of the women, 539 (12.3%) had mild or greater radiographic findings of OA of the hip in at least one hip, and 214 women (4.9%) had moderate to severe findings. Percentages of current and past users of oral estrogen were 17% and 24%, respectively. After adjustment for confounders, current use of HRT was associated with a significantly reduced risk of hip OA (OR 0.62, 95% CI 0.49–0.86). Current users who had taken HRT for more than 10 years had a greater reduction in the risk of OA of the hip (OR 0.57, 95% CI 0.40–0.82) compared with those using HRT for less than 10 years (OR 0.75, 95% CI 0.47–1.24). This sufficiently statistically powered study thus provides strong support for the notion that HRT protects against hip OA.
A subpopulation analysis of the Chingford study, including 606 postmenopausal women 45–63 years old, investigated associations between HRT use and OA prevalence at the knees and hands [34]. Standard anteroposterior radiographs of hands and knees were taken and scored according to both the methods of Kellgren and Lawrence (grade 2+ positive for OA) and individual features of osteophytes and joint space narrowing. Analysis compared ever use of HRT (> 12 months, n = 129) versus never use, and current use (> 12 months, n = 72) versus never use (n = 874). For current users, HRT was associated with a significant decrease in the risk for knee OA (OR 0.31, 95% CI 0.11–0.93). In addition, there was a notable trend for protective effects for moderate joint space narrowing of the knee and for distal interphalangeal OA. Taken together with results of the aforementioned study, the benefits of HRT are most pronounced at large joints in which articular cartilage is most challenged by weight bearing.
Prospective studies/analyses
There are relatively few long-term prospective studies that assess the impact of HRT on the incidence of radiographic OA. In an early analysis from the Framingham study, including 831 female participants aged 63–93 years, subjects who reported HRT use for at least 4 years had an apparent decreased risk for both radiographic OA (OR 0.71, 95% CI 0.42–1.20) and severe radiographic OA of the knee (OR 0.66, 95% CI 0.33–1.32) [35]. Later Zhang and colleagues revisited the same question in the Framingham cohort including women followed for 8 years [36]. Participants were 551 women who underwent radiographic examination of the knee at both baseline and at end of the follow-up. Global radiographic knee OA was assessed on a scale from 0 to 4. Individual radiographic features (osteophytes and joint space narrowing) were scored on a scale from 0 to 3. Worsening was defined as either development of radiographic OA that was not present at baseline (incident OA) or progression of baseline radiographic OA by ≥1 (progressive OA). When both incident and progressive radiographic knee OA cases were combined, current ERT use was associated with a 60% decrease in risk compared with never use (OR 0.4, 95% CI 0.1–1.5).
A prospective analysis from the Chingford study also provides important clues as to why estrogen deficiency is an important risk factor for progressive OA and why HRT is likely to provide countering effects [37]. From the original study population of 1003 women, those who were not receiving any bone-modifying medication, had radiographic information on the 4-year progression of knee OA, and had baseline information on bone mineral density were identified and separated into 4 groups as follows: controls (n = 50), progressive OA (n = 71), nonprogressive OA (n = 36), and osteoporosis (n = 59). The results demonstrated that bone resorption was increased in patients with progressive knee OA and was essentially unaltered in those with nonprogressive knee OA. The increase in bone resorption seen in patients with progressive knee OA was similar to that observed in patients with osteoporosis. These observations suggest that: the well-established implications of estrogen deficiency for accelerated bone turnover and related bone loss may also have implications for cartilage homeostasis; the beneficial effects of HRT in terms of inhibiting osteoclastic bone resorption might also provide therapeutic effects for the maintenance of cartilage integrity, if initiated in the early years of the menopause, when the impact of estrogen deficiency on bone loss is most pronounced.
Direct visualization of articular cartilage and measurements of its morphometric characteristics using magnetic resonance imaging (MRI) has only recently been standardized for prospective studies. A recent study by Wluka and colleagues elegantly demonstrated that women using HRT for at least 5 years had more articular cartilage in the knee than nonusers [38]. Although more and larger studies are needed to confirm these preliminary observations, the study is the first in a new avenue in the exploration of direct chondroprotective effects.
Conclusions
Collectively, current research on estrogen and OA appears to be conflicting, yet there is enough experimental and clinical evidence to warrant further investigation of this relationship. Although results of numerous large-scale studies suggest favorable associations between long-term HRT use and decreased risk of OA in weight-bearing joints, these associations are not conclusive. On one hand, the apparent chondroprotective effects might be confounded by the so-called ‘healthy woman effect’, which means that those women who choose to take hormone therapy may be more highly educated and more likely to lead a healthier life-style. On the other hand, it is also possible that HRT users in these observational studies represent women who initiated hormone therapy because they were in the need of this therapeutic option (e.g., for control of menopausal symptoms). This also means that they initiated their HRT in the early phase of their menopause, which could be a therapeutic window to maximize the benefits in terms of preventing some of the long-term consequences of the menopause. Extrapolating from results indicating the rapid acceleration of bone turnover and consequent bone loss in these years, mechanisms promoting cartilage degradation attributable to estrogen deficiency are most active in these years. Therefore, to obtain evidence-based conclusions as to the utility of HRT for the prevention of OA, we need randomized clinical trials including early postmenopausal women.
Although results of the Women's Health Initiative (WHI) study argue for no apparent relevance of HRT for primary prevention, the chondroprotective potentials of HRT were not included in the equation of the global index of risk and benefits that drove this conclusion [39,40]. Furthermore, there remains a shadow of doubt as to whether the conclusions of the WHI study arguing for potential adverse effects of estrogen plus progestin therapy are equally applicable to early postmenopausal women, who are the most likely candidates to benefit from the therapy in skeletal, cardiovascular, and quality of life terms.
Future perspective
The challenges of future trials will be the optimization of the estrogen dose and the better understanding of the parallel effects of progestins, which might lead to the optimization of the therapy for the estrogen to progestin ratio. In light of the promising findings of preclinical studies, evaluation of currently available and candidate SERMs might also provide us with effective alternatives to HRT for those who cannot or will not take HRT. However, before initiating these clinical trials, we should enhance the technical reliability of the methodological approaches and achieve consensus as to the standard set-up to be used for clinical assessment of chondroprotective agents. In this context much hope regards the new MRI technique, which has noteworthy potential to improve the sensitivity, accuracy, and reproducibility of cartilage assessment compared with the currently used radiography. MRI, in contrast to radiography, allows direct 3D visualization and quantification of articular cartilage. Furthermore, it can provide us with not only one joint space measurement approximating cartilage thickness, but a wide array of quantitative parameters (cartilage volume, cartilage thickness at numerous points, and joint surface area), and thereby a more detailed anatomical picture of articular status [41]. Even though the better imaging and 3D assessment of articular cartilage will likely improve direct monitoring of the structural changes of articular cartilage, the annual rate of changes and related statistical issues will still demand time-consuming trials and the inclusion of a large number of patients. Therefore, much expectation regards the new biomarkers of cartilage turnover that can signal dose-dependent deceleration of cartilage turnover far before structural changes emerge [42]. Of the various markers that have been proposed so far, degradation products of collagen Type II are the best characterized in terms of their ability to differentiate patients with OA from healthy individuals and predict the long-term progression rate of OA. Given the fairly small changes in cartilage thickness/volume when measured with imaging techniques, these biomarkers might prove to offer advantages in the identification of the optimal dose range of candidate drugs for Phase III trials.
It is the authors' belief that randomized, double-blind, placebo-controlled trials, which will use this combined methodological approach, will likely bring us closer to defining the role of HRT in the prevention of OA in early postmenopausal women.
Executive summary
Osteoarthritis (OA) is the most common joint disorder in the world. In the USA, it is second only to ischemic heart disease as cause of work disability in people over 50 years of age. Particularly after the menopause, women are more likely than men to get generalized OA, and interestingly, women have a tendency to develop a painful and deforming form of OA in the hand labelled ‘menopausal arthritis’.
Numerous animal models from rodents to nonhuman primates demonstrate the reciprocal effects of ovariectomy and ERT on cartilage turnover and cartilage degradation.
While the largest and statistically well-powered clinical studies provide supporting evidence for the beneficial effects of HRT for the prevention of OA, when taking all studies into account, this finding is not uniform. Differences in the estrogen plus progestin combinations, number of participants, as well as differences in methodological approaches and selected outcome variables are likely contributors to the diversity of findings.
Carefully designed randomized clinical trials including early postmenopausal women are needed to define the role of HRT. A standardized methodological approach combining advanced imaging techniques (MRI) and biochemical markers is expected to improve the sensitivity, accuracy, and reproducibility of cartilage measurements, which are the sine qua non of evidence-based evaluation of the chondroprotective effects of candidate drugs.