Risks and Benefits of Bisphosphonate Therapies

SKELETAL EFFECTS

The main skeletal benefit of bisphosphonates is the prevention of fragility fractures, which are a great socio-economic burden to health care systems. Fracture risk reduction with bisphosphonates in primary or postmenopausal osteoporosis has been proven in clinical trials; alendronate was found to reduce 50% of vertebral fractures in subjects with a previous vertebral fracture or with low bone mineral density, risedronate proved to reduce the vertebral, non-vertebral fractures (40–50% and 30–36%, respectively), and hip fractures (only in women aged 70–79 years old) and ibandronate was found to reduce the risk of vertebral fractures by 50–60% and the non-vertebral fractures only in a high-risk population (bone mineral density <−3SD) [Kanis et al., 2013]. Nevertheless, the increasing number of approved medications raises doubts as to which ones to use. Prioritizing one bisphosphonate over another is essential for decision-making in clinical practice, and this can only be achieved through the head-to-head trials. Ideally, these would have to include all of the available bisphosphonates in the market but unfortunately these kinds of trials are not feasible for practical and economical reasons. The mixed treatment comparison (MTC) studies can partially solve this problem.

MTC studies are an extension of meta-analysis, where analysis of three or more interventions can be carried out simultaneously in one meta-analysis. MTC studies can be used to analyze studies with multiple intervention groups by synthesizing direct and indirect comparisons, in the absence of head-to-head trials.

Although based on indirect estimates (only including placebo-controlled trials), MTC studies allow us to report multiple comparisons between bisphosphonates regarding fracture prevention. These studies found that zoledronic acid provided a greater vertebral fracture reduction (40% lower number of fractures) compared with ibandronate, alendronate, and risedronate [Jansen et al., 2009]. Furthermore, in spite of the proven efficacy of non-vertebral fracture prevention with alendronate and zoledronic acid, risedronate was the only bisphosphonate that was found to reduce non-vertebral fractures using the indirect comparisons of the MTC studies [Freemantle et al., 2013].

Once initiating a treatment with bisphosphonates, a usual question that arises is how to monitor the effect. Is the absence of fracture a strong enough indicator to continue bisphosphonates or is the appearance of a new fracture a sign that the bisphosphonate is not working? Indirect surrogates of fracture such as the bone mineral density or the bone turnover markers have been proposed as indicators to help answer these questions but evidence is still missing and no general consensus has been achieved [Kanis et al., 2013].

The continued treatment with bisphosphonates has raised concerns of its long-term anti-fracture effects. This has been addressed in several extension studies; the FLEX (alendronate 5-year extension study after the pivotal trial FIT), the risedronate 7 years extension study, and the 6-year zoledronic extension study all reaffirmed their fracture risk reduction with continued bisphosophonate use. Intriguingly, the discontinuation of alendronate did not increase the risk of non-vertebral or morphometric vertebral fracture [Eriksen et al., 2014]. The ibandronic acid extension study, the only one where fractures were analyzed only as an adverse event and not as a main outcome, found that 8.5–11% of the subjects in the ibandronate arm reported fractures while on treatment [Eriksen et al., 2014]. These results lead us to think that the overall fracture reduction efficacy is sustained with long-term exposure for the majority of bisphosphonates.

Despite the fracture risk reduction demonstrated in the RCT and comparison studies, results in the real-life setting have shown to be somewhat less optimistic. The lower fracture risk reduction observed in observational studies with real-life patient's poses into question the external validity of the bisphosphonates clinical trials. One of the possible explanations is the poor adherence [Feldstein et al., 2009] of BPs in real-life settings. While clinical trials reported a very high adherence rates (needed in order to see any effect in fracture risk reduction [Feldstein et al., 2009]), a cohort study reported that only 44% of the subjects treated with BPs had MPR > 80%, which could explain the decreased anti-fracture effect [Feldstein et al., 2009]. Accordingly, Danish national observational study found a risk reduction in hip fractures of 40% (HR 0.60; 95% CI 0.53–0.68) in patients with a cumulative use of five dose years of alendronate, compared with patients with more limited use [Abrahamsen et al., 2010].

Bisphosphonates have also proven their efficacy in other types of osteoporosis including glucocorticoid-induced osteoporosis. Glucocorticoids are frequently used in primary and secondary healthcare services. Glucocorticoid users are 1.6 times more likely to have a vertebral fracture and 2.6 times more likely to have a hip fracture with the risk of fragility fracture increasing mostly in the first 3 months after the glucocorticoid initiation [Briot and Roux, 2015]. This type of induced-osteoporosis is not merely linked to high dosages of glucocorticoids; studies have reported that even 2.5 mg per day may increase bone fragility [Briot and Roux, 2015]. Despite the increased awareness of healthcare professionals detecting and treating this disorder, only 25% of the subjects at risk receive treatment for their glucocorticoid-induced osteoporosis [Majumdar et al., 1998]. Treatment for this disorder starts with an early detection and initiation of BPs. These drugs have already proven their anti-fracture efficacy in osteoporosis but there is a paucity of trials with a fracture as a main outcome in glucocorticoids users. Among the BPs that have proven to increase BMD and are recommended to prevent bone loss among glucocorticoid users are alendronate, risedronate, and zoledronic acid [Briot and Roux, 2015].

Other bone-related diseases also benefit from the bisphosphonates anti-resorptive action. Bisphosphonates have proven to be effective and cost-effective in patients with rheumatoid arthritis [Van Staa et al., 2007]. However, the major advances have been achieved in Pagets' disease where bisphosphonates have proven a remission of the illness for long time periods [Reid, 2012]. Zoledronic acid has proven efficacy in reducing the risk of adverse skeletal events, including bone loss, in men with prostate cancer or in women with breast cancer [Coleman et al., 2014].

Another benefit of bisphosphonates, besides the inhibition of bone resorption and fracture risk prevention, is the relief of bone pain. The relief of bone pain occurs not only in rare bone diseases such as the fibrous dysplasia but also in bone cancer and bone metastasis. Bisphosphonates act as co-adjuvants with radiotherapy in order to decrease bone pain especially in those subjects where this pain is poorly localized. Bisphosphonates have proven to reduce by 50% the frequency of skeleton-related events in breast cancer and to reduce bone pain in prostate cancer. As a result, the ESMO (European Society of Medical Oncology) stated recently in their recommendations that zoledronic acids should be used in patients with metastatic breast or prostate cancer, and in selected individuals with lung, renal, and other solid tumor with metastasis [Coleman et al., 2014].

The main mechanism through which bisphosphonates act is suppression of bone resorption. All BPs have an affinity for bone tissue, and more specifically to osteoclasts because during bone resorption, the acidic PH of the resorption lacuna of the osteoclasts causes an intracellular uptake of BPs, leading to the internalization of substantial amounts of these drugs. Independently of the pathway used to inhibit bone resorption, either inducing a cytotoxic effect in osteoclasts or inhibiting the production of farnesyl diphosphate, all of them lead to the apoptosis of osteoclasts [Russell et al., 2008].

Overall, the benefits of bisphosphonates have been proven not only in primary and postmenopausal osteoporosis, but also in other bone-related diseases. The limited evidence of head-to-head trials makes it difficult to recommend one bisphosphonate over another since all of them have their indications and contraindications, so the choice must remain in its cost-effectivity and the type of patient for which the BP is been prescribed. If we consider the main indication of bisphosphonates, which is the fracture prevention in osteoporosis, some guidelines recommend starting with alendronate [National Institute for Health and Care Excellence, 2008) and risedronate [Mendoza et al., 2013] because of their high cost-effectivity [Compston et al., 2009].

EXTRA-SKELETAL EFFECTS

The main extra-skeletal benefit of bisphosphonates identified in RCTs is the finding of a reduction in mortality in a hip fracture patients randomized to zoledronic acid. In the placebo group, 141 of 1,057 patients died during the study compared with 101 of 1,054 patients in the zoledronic acid treated arm, a risk reduction of 28%, chiefly explained by fewer deaths from cardiovascular causes [Lyles et al., 2007], though the mechanism is not clear. In rats, knock-down of farnesyl pyrophosphate synthase protects against hypertensive cardiac hypertrophy [Ye et al., 2010] and zoledronic acid could have a positive effect in some animal models of accelerated ageing. A number of observational studies have linked BP use to better survival but this could be due to elderly patients in poor health being potentially less likely to begin BP treatment. However, if BPs truly extend life expectancy—an outcome that clinical trials were not designed to or necessarily powered to pursue—it would not make sense for a high risk of mortality to be a barrier for treatment, rather the opposite.

Some observational studies, which should be regarded as hypothesis generating rather than proof of effect, have also suggested that oral bisphosphonates could reduce the risk of colon cancer and possibly gastric cancer [Abrahamsen et al., 2011; Pazianas et al., 2012; Singh et al., 2013]. Though a reduced risk of breast cancer was also suggested by the Women's Health Initiative [Chlebowski et al., 2010] (observational design), this was recently refuted by analysis of RCT data from the primary licensing trials of alendronate [Hue et al., 2014]. Finally, a twofold increased implant survival of the total hip or knee prosthesis was detected in a recent cohort study carried out in the United Kingdom. The protective effect was found to be greater in subjects with a knee rather than a hip prosthesis [Prieto-Alhambra et al., 2011].

SKELETAL POTENTIAL HARM

Where early bisphosphonates had the capacity to inhibit bone mineralization and induce osteomalacia at high doses, the skeletal concerns with modern nitrogen-containing bisphosphonates (sometimes inappropriately referred to as aminobisphosphonates) relate to lowering of bone resorption itself. Lower bone resorption may indirectly reduce the rate at which bone microdamage is generated but it will also directly reduce the rate at which microdamage is removed and replaced by new bone. Atypical femur fractures (AFF) may be an indication that artificially induced low bone turnover over years may lead to unexpected novel fractures. The term atypical femur fracture refers to a rare group of femur fractures that have specific radiological features, which include a pronounced predilection for the lateral aspect of the subtrochanteric or diaphyseal femur, a mainly transverse fracture direction and a tendency for periosteal or endosteal thickening of the cortex. The diagnostic criteria are given in detail elsewhere [Shane et al., 2014]. The fractures may be bilateral and incomplete fractures often show delayed healing and may need surgical pinning. While such fractures are much more common in patients treated with bisphosphonates they do also occur, albeit rarely, in the background population.

Data from Kaiser California in the United States reported an age-adjusted incidence rate of 1.8 per 100,000 with up to 2 years of BP use, 16 per 100,000 with 4–6 years of use, and 107.5 per 100,000 with more than 10 years of use, suggesting an exponential increase in risk with increasing duration of use [Dell et al., 2012]. An exponential dose–response relationship has also been reported by Swedish researchers who first published findings based on reviewing all potential AFF X-rays in one calendar year for the Swedish nation [Schilcher et al., 2011]. More recently, the review period has been extended to 3 years. The updated analysis finds a relative risk of AFFs in Swedish BP users of more than 4 years of exposure of 126 (95%CI 55–288) or 110 excess fractures per 100,000 patient years. This rate is six times higher than expected from Kaiser California [Dell et al., 2012]. It is not however impossible as the rate falls within the 440 per 100,000 total rate of subtrochanteric and shaft fractures (of which they are a subgroup) reported in Danish women with an average of 4 years of BP use, a rate which declined to 320 per 100,000 with 9 years of use [Abrahamsen et al., 2010]. However, weaknesses of the Swedish dataset include the short prescription history provided by the national register, which could underestimate duration of BP use and inflate the dose–response association. The potential for harm should be balanced against the much higher rates of typical osteoporotic fractures; hence, Swedish women treated with BPs experience hip fractures at a rate of 1,500 per 100,000 [Schilcher et al., 2011] and are also at similarly high risk of other major osteoporotic fractures. It is unclear if all long-term users of BPs are at risk of AFF or if this risk is confined to a distinct subgroup of patients; at present, the risk of AFFs appears substantially higher in persons of East Asian heritage than in Europeans. Vitamin D deficiency, diabetes, and proton pump inhibitor use have been proposed as potential risk factors for AFF. It is unclear at present if the risk of AFF is lower with bisphosphonates with a shorter residence time in bone such as risedronate, the majority of observations having been made in alendronate users.

Among patients treated for osteoporosis, osteonecrosis of the jaw (ONJ) occurs at a much lower rate than AFF but ONJ is not rare when bisphosphonates are used in an oncology setting. The condition has been known for many decades in other contexts. It was first described in factory workers exposed to large amounts of phosphorous in the manufacture of matches and since identified as a consequence of local radiation therapy. Though the necrotic process may not initially be exposed to the naked eye, the working definition of ONJ remains a clinical diagnosis based on identification of an area of exposed bone in the oral cavity that fails to heal in 8 weeks [Khan et al., 2015]. An earlier stage, unfortunately termed non-exposed ONJ, has been proposed to cover cases of ONJ diagnosed by radiological imaging alone [Schiodt et al., 2014]. The epidemiological data for ONJ in patients treated with BPs for osteoporosis were recently reviewed by Solomon et al. [2013], finding broad variation—of a hundredfold or more—in reported incidence rates and relative risks between studies. The authors then analyzed two claims databases and found one confirmed case per 4,900 BP users in one cohort and one suspected case per 14,300 BP users in their other cohort, producing cross-sectional prevalence estimates that fell within the range reported in the prior literature. There is no clear indication that the risk of ONJ increases with exposure time with osteoporosis doses. In oncology, the risk of ONJ seems much higher than in osteoporosis, reflecting probably not only the shorter dosing interval (larger annual cumulative exposure) but also the effects of anti-neoplastic agents such as angiogenesis inhibitors, immune suppression, and prednisolone treatment. The highest cross-sectional prevalence of ONJ presently reported in the oncology field is eight cases out of a study population of only 43 patients with advanced prostate cancer who had extensive dental examinations. Incidence estimates [Khan et al., 2015] for ONJ with i.v. bisphosphonates in oncology have ranged from 0 to 12,200 per 100,000 patient years (so up to one case per eight patient years). There remains a substantial lack of solid data on the risk factors for ONJ, the absolute risk as a function of time and dose, and not least the underlying pathophysiology including the relative role of vascular, epithelial, and skeletal events and the sequence in which they develop and mature into clinically evident ONJ. Management of ONJ is beyond the scope of this review but readers are referred to the consensus paper by Khan et al. [2015].

EXTRA-SKELETAL SIDE EFFECTS