Introduction

Osteoarthritis (OA) is a common chronic disorder characterized by joint cartilage wearing as the central feature, associated with concomitant change in the synovial and subchondral bone metabolism [1]. Knee osteoarthritis (KOA) is a major cause of disability affecting about 10 % of people over the age of 55 years [2]. Current treatments aim to alleviate both pain and functional disability by a combination of non-pharmacological and pharmacological approaches, while the efficacy of disease modifying drugs is controversial [3, 4].

The pathogenesis of OA remains a subject of controversy [5]. OA initially presents mostly with destruction of the articular cartilage [6], which is followed by osteophyte formation, cystic degeneration and then sclerosis of the subchondral bone [7]. This “cartilage first” hypothesis has been challenged, and it was hypothesized that also changes in subchondral bone may play an important pathophysiological role [8]. These changes include the subchondral cortical plate and the underlying cancellous bone. The subchondral bone in OA has been shown to have a decreased number and thinning of cancellous trabeculae and localized osteoporosis [9, 10], but also increased bone mineral density (BMD) as assessed by dual X-ray absorptiometry [11, 12], and increased bone turnover, micro-fractures, and bone marrow lesions on magnetic resonance imaging (MRI) [13–15]. A recent systematic review on MRI abnormalities in KOA concluded that knee pain is associated with bone marrow lesions and effusion [16].

Therefore, subchondral bone might be a key target for OA treatment [17]. In particular, subchondral bone turnover and osteoclast functions are likely to be important elements in the pathogenesis of OA [17]. In patients with KOA, fewer MRI subchondral lesions after treatment with drugs that suppress bone turnover, including bisphosphonates, have been observed [18]. Bisphosphonate therapy was found to decrease the incidence and progression of osteophytes in animal models [19–22]. Moreover, a direct chondro-protective effect of bisphosphonates has been postulated [23]. However, bisphosphonate use has been explored in a number of studies on patients with spondylosis and/or KOA with inconclusive results. An improvement in subjective pain was reported in one uncontrolled study of etidronate [24], and the treatment with alendronate was associated with less spinal osteophytes and disk-space narrowing progression than placebo in a post hoc analysis of fracture intervention trial [25]. A treatment with risedronate was associated with a trend of pain relief, improvements in patient global assessment of disease scores and progress in delaying radiographic progression in patients with KOA in a pilot study [26], but these results were not confirmed in a properly powered study [27]. Zoledronic acid has been reported to reduce subchondral bone marrow lesions and knee pain [28]. Recently, data from NIH osteoarthritis initiative strengthens the concept that treatment with bisphosphonates may be associated with beneficial symptomatic effects and perhaps structural benefits in people with OA [29].

A major limitation in the studies assessing the efficacy of bisphosphonates in OA is the need of high doses with potential safety concerns [30], which are largely taken up by bone tissue and therefore, poorly concentrated on subchondral bone. An alternative approach is represented by the administration of intra-articular (IA) bisphosphonate.

In a prior multicenter study, IA clodronate was apparently as efficacious as hyaluronic acid, with a significant progressive improvement in pain and functional outcomes with both drugs extending for at least 2–4 weeks after the last injection [31]. IA hyaluronic acid is recommended by prestigious scientific societies [3, 4, 32], but long-term efficacy is still highly controversial. A recent systematic review and meta-analysis of 89 trials on IA visco-supplementation of the knee concluded that the benefit on pain and function is minimal or absent with a high risk of flares up [33]. These conclusions may represent a relevant draw back for our original study with IA clodronate, persisting the possibility that both IA clodronate and hyaluronic acid had exclusively a relevant “placebo” effect.

The “placebo” effect of saline injection has been described in particular in the first weeks after IA injection [34–37]. For this reason with this study, we wanted to compare in a double-blind design the tolerability and efficacy of IA clodronate versus saline solution after 5-week period after the last injection.

Materials and methods

This is a randomized, single-center, double-blind phase 3 study aimed to evaluate efficacy and tolerability of IA clodronate 2 mg versus placebo in KOA.

Men and non-pregnant women aged between 50 and 75 years, suffering from KOA defined according to American College of Rheumatology criteria [38], radiographically confirmed with a Kellgren–Lawrence scale [39], with X-rays performed in the last 6 months, symptomatic for at least 3 months, were eligible. The Kellgren and Lawrence system is a method of classifying the severity of knee osteoarthritis on a five grade 0–4 scale : grade 0—no radiographic features of OA are present; grade 1—doubtful joint space narrowing (JSN) and possible osteophytic lipping; grade 2—the presence of definite osteophytes and possible JSN on anteroposterior weight-bearing radiograph; grade 3—multiple osteophytes, definite JSN, sclerosis, possible bony deformity; grade 4–large osteophytes, marked JSN, severe sclerosis, and definitely bony deformity. Additional inclusion criteria were baseline spontaneous pain not <40 mm on visual analogue score (VAS) scale of 100 mm and reduced joint mobility greater than grade 2 on a 4-point scale (arbitrarily assessed by the investigator using a scale from 1 complete to 4 totally compromised).

Patients were excluded if Kellgren–Lawrence score was 1, if they had osteoarthritis secondary to other joint diseases, if they were receiving anticoagulant therapy or if they had received systemic steroidal drugs, chondro-protective agents, or intra-articular drugs in the 30 days preceding the study. Patients were also excluded if they had been on bisphosphonates in the previous year or if they had undergone knee surgery in the past (excluding arthroscopy) or washing, diagnostic or surgical arthroscopy in the previous 180 days. Patients with known hypersensitivity to study drugs or similar drugs, including acetaminophen or suffering from other diseases that require use of anticoagulants, were excluded. Individuals coming from outside our province and then unlikely to be able to attend all the visits and those with a history or suspected alcoholism or drug addiction, patients with poor compliance or unable to provide informed consent were also excluded.

The primary objective was the sum of spontaneous, on passive movement, and at digital pressure pain (4 kg/m2) [40] relief of the index knee, all evaluated by VAS of 0–100 mm reported by patient at clinical evaluation, at week 8 (5 weeks after the last injection) in order to minimize the expected placebo effect of the injections. We use the sum of different aspects of pain to assure a more global assessment of joint pain.

Improving in Western Ontario MacMaster (WOMAC) scale, Lequesne index, consumption of acetaminophen, and physician or patient overall judgment were secondary objectives.

The WOMAC Index [41] is a three-dimensional, self-administered measure specific to OA, looking at pain, stiffness and physical function in OA of hip or knee. There are 24 questions, five on pain, two on stiffness and 17 on physical function.

The Lequesne-algo-functional index for knee [42], looking at pain, maximum distance walked and activities of daily living. Questions are asked and points allocated according to the patient’s reply on a 3-point Likert scale (0–2).

The consumption of acetaminophen tablets was reported at each visit in total number/week (counting the returned pills that were provided weekly by the investigators); for simplification, we later divided patients in requiring or not the rescue drug.

Outpatients who attended our clinic complaining of knee pain due to KOA, and meeting the inclusion criteria and without exclusion ones, were proposed to take part to the study. At baseline, eligible patients without any stratification were randomized at a 1:1 ratio for blocks of 4, using a randomization list generated with computer software to receive either once weekly IA injection of 2 mg clodronate or placebo for 4 weeks with further 6 fortnightly follow-up visits. The injected volume was 1 ml. The vials containing clodronate or placebo were prepared at the Abiogen Pharma (Pisa SpA, Italy) laboratories, where stability and the final concentration of clodronate were periodically tested. The treatment codes were kept by Abiogen Pharma, and they were disclosed after the registration of the locked database at the competent office of Ministry of Health. The placebo and active vials were totally indistinguishable. If both knees met the inclusion criteria, only the worst received the treatment to avoid the bias due to the certain influence of a worse knee taking the better as study target. Four days before the first injection patients were asked to discontinue any nonsteroidal anti-inflammatory drugs, and physical therapy and rehabilitation procedures were not permitted over the entire follow-up. During the study (i.e., the 4 days preceding the first injection and afterward), patients were allowed pain relief with acetaminophen as necessary.

The enrolled subject could discontinue the study at any time on his own initiative, without prejudice to the possibility to make use of other therapies. The date and reasons for discontinuation were reported in the appropriate case report form, indicating one of the following reasons: (1) adverse event; (2) serious adverse event; (3) therapeutic effect is not satisfactory; (4) the appearance of the exclusion criteria; (5) uncooperative patient; (6) withdrawal of consent; and (7) administrative problems.

The safety population comprised all patients receiving at least one injection of the study medication. Adverse events observed by the investigators or reported by the patients spontaneously or following a non-leading question were coded using the medical dictionary for regulatory activities. Particular attention was paid to local painful reactions at the injection site, postinjection reactions (e.g., effusions), and acute pseudoseptic arthritis.

The clinical study was conducted in accordance with the ethics principles of the Declaration of Helsinki and was approved by the local ethics committees. Each enrolled patient signed the informed consent to participate in the study.

The trial is registered in the National Observatory on clinical investigation of medicinal: EUDRACT 2009-012956-26.

Sample size and statistical analysis

The sample size of 40 patients per treatment group, assuming a retention rate of 85 %, has 90 % power, with a 0.05 two-sided significance level, to detect a difference between treatments of 20 mm in the cumulative VAS value, assuming that the common standard deviation is 24 mm according with our previous experience [31]. Demographic and main clinical characteristics of the treatment and placebo groups were summarized by descriptive statistics. Treatment groups have been compared at each visit carrying out a mixed model for repeated measures, using the change from baseline as dependent variable, treatment, visit, and the treatment by visit interaction as fixed factors, baseline as covariate, and patient as random effect [43]. Differences between treatments have been reported as least-square mean estimates. Intention-to-treat (ITT) analysis, conducted using data from all randomized patients who received at least one dose of study drug and the last value carried forward procedure, was included. Statistical significance was determined using a p value ≤0.05 (two tailed).

All analyses were performed with SPSS, version 13.0 (SPSS Inc., Chicago, IL, USA).

Results

Eighty patients, 67 women and 13 men aged 66 ± 6 (SD) years, were recruited from the outpatients clinic with the service for intra-articular injections. Nine screened patients were excluded because not fulfilling the inclusion or exclusion criteria. This service is fully reimbursed by the regional health service, and this might explain why all patients offered to participate in the trial gave their consent. Two patients drop out: one on clodronate after the second injection for family problems and the second (on placebo) during 12-week follow-up for an adverse event unrelated to treatment (Fig. 1). There were no additional missing values. The main clinical characteristics of the study population are listed in Table 1. Only 12 patients had been on bisphosphonates and all of them discontinued the treatment at least 18 months before recruitment.

Fig. 1
figure 1

Flow diagram of recruited patients. F females, M males

Full size image
Table 1 Clinical characteristics at baseline of the patients on clodronate (8 men, 32 women) or placebo treatment (5 men, 35 women)
Full size table

A significant improvement for all efficacy parameters was observed at all-time points in both groups. The difference between the two groups in favor to clodronate for the primary end point of the sum of the three VAS pain scores at 8-week follow-up (5 weeks after the last injection) was 27.4 mm thus fulfilling the assumption made for the esteem of the statistical power. The difference between the two groups completely disappeared at 12-week follow-up, 9 weeks after the last injection (Table 2). The mean changes for the three individual VAS pain scores are also listed in Table 2.

Table 2 Changes in visual analogue score (VAS, means and SD) of spontaneous, on passive movement, at digital pressure pain, and their sum, at different time points
Full size table

The improvements in global KOA evaluation from both patient and investigator (Fig. 2), Lequesne index and WOMAC pain subscale, were significantly greater in clodronate treated patients at 8-week follow-up (5 weeks after the final injection) (Fig. 3), but the changes in global WOMAC score were not significantly different.

Fig. 2
figure 2

KOA evaluation from both physician and patient (mean ± SE) (down arrow time of intra-articular injection). *p < 0.05 between the two treatments

Full size image
Fig. 3
figure 3

Lequesne index (mean ± SE) and WOMAC pain subscore (mean ± SE) (down arrow time of intra-articular injection). *p < 0.05 between the 2 treatments

Full size image

Approximately half of the patients required acetaminophen treatment during the 4 days preceding the first injection. This proportion, together with the mean number of tablets (data not shown), remained stable in the placebo group while they progressively declined in the active group, particularly during the posttreatment follow-up (Fig. 4).

Fig. 4
figure 4

Percentage of patients taking acetaminophen in the two groups during the follow-up (down arrow time of intra-articular injection). *p < 0.05 between the two treatments

Full size image

The results of these analyses and the related statistical significances did not change when modified ITT analyses conducted using data from all randomized patients who received at least one dose of study drug and the last value carried forward procedure were adopted.

The mean reported pain at the site of IA injection was significantly higher in clodronate group (Fig. 5), but local pain never led to treatment discontinuation. No other side effects related to the treatment were reported by all patients, included the 2 who abandoned the trial in advance. Side effects, considered unrelated to treatment, were reported by 15 patients in the control group (flu-like syndromes, diarrhea, abdominal pain at the time of the infusions and flu-like syndromes, abdominal pain, low back pain, and headache during follow-up) and by 10 in the clodronate group (flu-like syndromes, abdominal pain at the time of infusions and low back pain, diarrhea, neck stiffness and in one case severe cough that led to hospitalization for suspected lung cancer during follow-up).

Fig. 5
figure 5

Tolerability: VAS pain at the site of IA injection in the two groups (down arrow time of intra-articular injection). *p < 0.05 between the two treatments

Full size image

Discussion

In this study, we have shown that the knee IA injection of 2 mg clodronate is associated with significantly greater benefits than the infusion of saline solution in a double-blind randomized clinical trial. A striking finding of our study is the impressive 35–45 % beneficial effect observed in both groups for the primary efficacy parameters throughout the study, with the two groups diverging only after the end of the fourth injection. A similar effect of saline injection has been reported by others [34–37], and it was attributed to the injection per se, aspiration of synovial fluid, “washing” effect of the saline solution, education of the patients, or placebo effect.

In our previous dose-finding study [31] with IA clodronate, we found a trend of dose–response relationship up to 2 mg clodronate and no significant differences between clodronate and hyaluronic acid. In that study, a placebo group was not included and the interpretation of the results could not be conclusive regarding the pharmacological efficacy of clodronate since an equivalent placebo effect could be attributed to both treatments. The efficacy of hyaluronic acid in KOA is still debated. The analysis from meta-analyses yielded conflicting results: Some suggest small to strong effects [44–47], but others failed to show an effect superior to placebo [48, 49] attributing the results of other analysis to the presence of publication bias that may have been responsible of an overestimation of the observed small effects [50]. The recent systematic review and meta-analysis [33] concluded that the benefit of visco-supplementation on pain and function is minimal or nonexistent.

The results of the present trial do indicate an additional small benefit of clodronate over saline solution. Most indices are slightly but significantly better and tend to affect both pain and functionality as expressed by WOMAC score and Lequesne index. The differences are limited to the first month after the last injection but tend to disappear thereafter.

Bisphosphonates may work through a variety of mechanisms, such as the suppression of local bone turnover or via inhibition of local pro-inflammatory cytokines or nociceptors. Defective subchondral bone metabolism in osteoarthritis could modify chondrocytes in cartilage via local factors released during subchondral bone remodeling. A direct stimulation of collagen and proteoglycan synthesis [19] and MMP inhibition and downregulation [23] have been also reported. TGF-β1 activated during osteoclastic bone resorption in subchondral bone seem to initiate the pathological changes of osteoarthritis [51] and inhibition of this process by bisphosphonates could be a potential therapeutic approach to treating this disease. Bisphosphonates can exert a mild immune-modulating activity via inhibition of pro-inflammatory cytokines [52] or white blood cells subtypes [53, 54]. Clodronate containing liposomes have been tested in a model of rheumatoid arthritis with evidence of synovial macrophage depletion [55]. Moreover, a recent study using IA zoledronic acid in a rat OA model reported a suppressive effect on synovial inflammation [56]. Increased subchondral bone turnover may contribute to pain in OA and targeting of osteoclasts may reduce pain associated with OA [57]. Bisphosphonates also have effects on cells other than osteoclasts, such as macrophages and microglia, including those that produce the pro-nociceptive β-NGF, which may explain their effect in algo-dystrophic syndromes [58]. Bisphosphonates modulate the release of protons from activated osteoclasts, and this may reduce the acidic sensitization of bone nociceptors [59], and a direct effect on the synovial glial cells has also been reported [60].

From secondary analysis of patients treated with bisphosphonates or strontium ranelate for osteoporosis, a trend of reduction of joint space loss has been reported [25, 26, 29, 61], but this was not confirmed in a larger ad hoc study with risedronate [27]. Some have speculated that this clinical trial was underpowered, but it is unlikely that new properly designed studies will be undertaken in the future, taking into consideration the newly arisen safety issues of long-term bisphosphonate treatment, attributed to the long skeletal retention and over-suppression of bone turnover.

Even though a direct evidence that the bisphosphonate may diffuse from cartilage to bone is missing, it is likely that the approach that we adopted with the IA administration of clodronate generates local high concentration of the drug on joint tissues and subchondral bone, thus circumventing the problem of systemic administration of bisphosphonates being largely taken up by the entire skeleton [62]. A direct effect of clodronate on cartilage cannot be ruled out since clodronate was shown to bind fibroblast growth factor 2 and polysulphated or polyphosphated derivatives [63] and to modify cartilage oligomeric matrix protein [64].

Whatever the mechanism of action, the bisphosphonate injected IA and then likely bound to sub-chondral bone is expected to exert its action for quite long time. The reason for the persistent effect of bisphosphonates is still very controversial with some attributing this to the bisphosphonates absorbed on the resting surfaces and others to re-circulation of the bisphosphonates buried in the bone tissue. In both cases, the persistence of the effect is related with the affinity for the mineralized tissue. Clodronate is by far the bisphosphonate with the lowest affinity [65] and its activity on bone turnover even after very high doses is expected to disappear within few days after discontinuation. In our previous study [31], even though insignificant, the best results were obtained with Clodronate 1 mg twice a week for 2 weeks rather than 2 mg once a week.

The main limitation of our present study is the limited benefit achieved with clodronate versus placebo and that makes our results still inconclusive regarding the possibility to develop a new therapeutic strategy for KOA. Other limitations are the lack of useful informations on structural modification effects, such as radiological, laboratory, or synovial histological data, but the study was aimed to clinical outcomes. The significant difference between the two treatments in the global pain scores persisted for a month after the last injection. The difference observed versus saline injection was somewhat lower than that assumed in our power calculation and this may raise concerns regarding the statistical power of our study. The substantial effects observed after each weekly injection of saline solution might have obscured the long-term superior efficacy of clodronate. In future studies, the possibility to test monthly injections of both saline solution and clodronate together with greater doses of clodronate are enviable, taking into consideration the good local tolerability of 2 mg. The real value of this study is to provide some additional support to the hypothesis that subchondral bone turnover plays a role in the development and maintenance of KOA.

In conclusion, we have seen that IA 2 mg clodronate is safe and is associated with small symptomatic and functional benefits in KOA patients in addition to those seen with saline solution, with benefits appreciable for a few weeks after the treatment course.