In clinical trials, both oral and intravenous bisphosphonates have been well tolerated [15]. Upper GI intolerance, usually mild or moderate, is observed with oral dosing in daily practice, although such intolerance was not observed in clinical trials [16]. The incidence of upper GI symptoms among patients receiving placebo in the oral bisphosphonate clinical trials was 30–50 %. Such a high background incidence might have blunted the ability to observe GI intolerance in the trials. Poor adherence to the oral bisphosphonate dosing regimen increases the likelihood of upper GI symptoms . It is unclear whether differences exist in the risk of upper GI intolerance among the oral bisphosphonates [17]. GI bleeding and esophageal ulceration or rupture have been described very rarely.

Acute phase reactions with fever and myalgia occur with intravenous or high dose oral therapy [18–21]. These symptoms occur in about 30 % of patients after the first intravenous dose of zoledronic acid in bisphosphonate-naïve patients but occur much less frequently after subsequent doses or in patients previously treated with oral bisphosphonates. Pretreatment with antipyretic medications reduces the incidence and intensity of symptoms [22]. Bone and muscle pain, not related to acute phase reaction, has been reported after oral or intravenous dosing. Since these symptoms were not observed in clinical trials, neither the incidence nor the mechanism of this potential side effect is known.

Transient decreases in serum calcium levels occur upon initiating treatment. Hypocalcemia can occur, especially in patients with renal insufficiency, osteomalacia, vitamin D deficiency, or hypoparathyroidism [23–27]. Ensuing adequate intakes of calcium and especially vitamin D prior to treatment is probably protective, and therapy is contraindicated in hypocalcemic patients.

Bisphosphonates are cleared from the circulation by either binding to the skeleton or by renal excretion. Rapid intravenous infusion of pamidronate or zoledronic acid to patients with cancer-related bone disease has been associated with renal failure due to focal glomerular sclerosis or acute tubular necrosis [28–30]. Based on this knowledge, patients with severely impaired renal function were not included in clinical osteoporosis trials with bisphosphonates. In the studies with oral agents, no effects on renal function were observed. Among the small numbers of patients with CDK3 renal function enrolled in those trials, no adverse effects on renal function or impaired efficacy were observed [31]. However, therapy with oral bisphosphonates is not recommended in patients with estimated GFR <30 cc/min. Individual cases of acute renal impairment related to oral bisphosphonates have been reported. In the pivotal HORIZON studies, transient increases in serum creatinine were observed after intravenous zoledronic acid, but over the course of the studies, the age-related decline in renal function was similar between treatment and placebo groups [32]. Acute renal failure and death after zoledronic acid infusion have been reported in clinical practice, although usually in patients with cancer-related bone diseases. Zoledronic acid is contraindicated in patients with a creatinine clearance of <35 cc/min [33].

Bisphosphonate nephrotoxicity is related to the maximal concentration of drug in the circulation perhaps explaining the paucity of adverse renal effects of oral bisphosphonates [30]. For patients with CDK3, extending the infusion time of zoledronic acid from the usual 15 to 30 min or even 60 min should limit the Cmax achieved with treatment and perhaps minimize the risk of renal injury.

Atrial fibrillation associated with hospitalization occurred more frequently (1.3 %) with intravenous zoledronic acid compared to placebo (0.5 %; p < 0.001) in the HORIZON Pivotal Fracture Trial [2]. Cardiac arrhythmia could theoretically be caused to the transient decrease in serum calcium after infusion. However, the cases of serious adverse events of atrial fibrillation were not temporally related to the annual dose. Furthermore, the incidence of atrial fibrillation, other arrhythmias, or cardiac events and stroke was similar between treatment and control groups. No association of zoledronic acid therapy and atrial fibrillation was observed in the HORIZON Recurrent Fracture Trial [34]. After review of all bisphosphonate studies, the FDA concluded that a causal link between bisphosphonate therapy and atrial fibrillation has not been established [35]. Driven primarily by the HORIZON Pivotal Fracture Trial, recent systematic reviews continue to suggest an increased risk of atrial fibrillation with intravenous bisphosphonates [36].

No other cardiovascular concerns were noted in clinical trials with bisphosphonates. Although observational studies have demonstrated an association between bisphosphonate therapy and decreased risks of hypertension, myocardial infarction, and stroke, recent meta-analyses do not suggest an association between bisphosphonate therapy and cardiovascular or cerebrovascular disease [37, 38].

Cases of esophageal cancer have been reported in patients who took oral bisphosphonates [39]. No evidence of this association was observed in placebo-controlled clinical trials or in several observational cohorts [40–42]. One of two analyses of the UK General Practice Research Database suggested an increased risk of esophageal cancer (3 vs. 2.3 % in controls) and especially with treatment for >5 years (RR 2.24, 1.47–3.43) [42]. No relationship between bisphosphonate use and esophageal cancer was observed in the other analysis of that same database [43]. The FDA has stated that evidence is insufficient to evaluate this association but recommends that oral bisphosphonates not be used in patients at risk for esophageal cancer including those with Barrett’s esophagus [44].

In contrast to the associations of bisphosphonate therapy with adverse outcomes, decreased risks of breast, prostate, colon, and pancreatic cancer have been reported with bisphosphonates in observational studies. However, a recent analysis of pooled randomized clinical trials found no association between bisphosphonate therapy and breast cancer over an average follow-up of 3 years.

Mortality was reduced by 28 % with IV zoledronic acid therapy in the HORIZON Recurrent Fracture Trial [1]. Decreased mortality in patients treated with bisphosphonates has also been noted in observational studies [45–48]. These effects on mortality were greater than could be explained by the anti-fracture effects of therapy.

Rare cases of inflammatory eye disease (uveitis, iritis) have been described in patients receiving oral and IV bisphosphonates including older non-nitrogen-containing agents like etidronate and tiludronate [49, 50]. In a large register-based cohort, the incidence of hospital-treated uveitis was very low (0.05 %) during the first 12 months of prescription therapy for osteoporosis, with no difference observed between patients receiving bisphosphonates or other osteoporosis drugs [51]. Among 1001 patients receiving intravenous zoledronic acid in a clinical trial, eight (0.8 %) developed acute uveitis within the first few days of treatment [52]. Very rare cases of hypersensitivity and anaphylaxis have been reported with bisphosphonate therapy [53, 54].

In clinical trials, these agents are generally well tolerated [58, 59]. Vasomotor symptoms and muscle cramps , usually of mild to moderate severity, occurred more frequently with each drug compared to placebo.

Estrogen-like increases in the risk of venous thrombotic events (VTE) have been reported with each drug. In the pivotal phase 3 MORE study, the incidence of VTE was increased in patients receiving raloxifene 60 or 120 mg daily for up to 40 months (1.0 %) versus the placebo group (0.3 %; RR 3.1; 95 % CI 1.5–6.2) [55]. The greatest risk occured within the first 4 months of therapy and did not appear to increase with prolonged exposure to the drug during years 5–8 of treatment with raloxifene 60 mg daily (CORE Study) [56]. Over 8 years of follow-up, the incidence rates for venous thromboembolic events were 2.2 and 1.3 events per 1000 woman-years for the raloxifene and placebo groups, respectively. Over 3 years of therapy with bazedoxifene 20 mg daily, the rate per 1000 woman-years was 2.86 versus 1.76 in the placebo group [57]. As with raloxifene, the rate of VTE was highest in the first year (relative risk of 2.69). During the 5 years of the PEARL study, the rate of VTE with lasofoxifene 0.5 mg daily was 2.9/1000 patient-years compared to 1.4 with placebo [HR 2.06 (95 % CI 1.17–3.61)] [58]. These agents are contraindicated in women with a prior history of venous thrombotic event and are to be used with caution in women at risk for VTE. Temporarily stopping treatment is recommended prior to and during prolonged immobilization [60].

No differences in the incidence of cardiovascular or cerebrovascular events, in complications from those events or in mortality, were observed during the 8 years of the MORE and CORE studies [61, 62]. The RUTH study evaluated the effect of raloxifene versus placebo in postmenopausal women over a median duration of follow-up of 5.6 years in more than 10,000 postmenopausal women, average age 67.6 years, with documented coronary heart disease or at increased risk for coronary events [63]. Although stroke and cardiovascular events occurred similarly in women who received raloxifene 60 mg daily or placebo, death related to stroke did occur more commonly with raloxifene (1.2 %) compared to placebo (0.8 %) (HR 1.49, 95 % CI 1.00–2.24; p = 0.0499), an increase from 15 to 22 per 10,000 woman-years. Raloxifene had no significant effect on all-cause mortality [64]. Lasofoxifene 0.5 mg daily was associated with lower rates of myocardial infarction and stroke [56, 65]. Rates of fatal stroke (HR 1.40; 95 % CI 0.44–4.4) and overall mortality (HR 1.12; 95 % CI 0.80–1.56) did not differ significantly between the lasofoxifene 0.5 mg daily and placebo groups. There were no significant differences in the incidence of cardiovascular events or stroke between patients treated with bazedoxifene or placebo [66].

In studies of postmenopausal women with osteoporosis and the RUTH trial, raloxifene has consistently decreased the risk of invasive breast cancer with overall reductions in relative risk of 44–69 % [67–69]. The relative reduction in risk with raloxifene appears to be similar across the spectrum of risks in the various patient groups. In the STAR study, protection from invasive breast cancer was similar in women at high risk for breast cancer receiving tamoxifen or raloxifene [70]. Treatment with lasofoxifene 0.5 mg daily for 5 years reduced the risk of invasive breast cancer by 85 % (95 % CI 50–96 %), and the effect was greater in women with serum estrogen levels above the group mean [71]. These effects are limited to the risk of invasive estrogen receptor-positive breast cancer [72]. No effect of bazedoxifene on breast cancer risk has been reported.

Raloxifene had no effect on cognitive function [73]. Over 5 years, no endometrial safety issues were observed with raloxifene [74]. Compared to placebo, lasofoxifene therapy for 5 years resulted in small increases in frequency of vaginal bleeding, endometrial thickness, and vaginal polyps but no difference in endometrial cancer or gynecological surgery for uterine prolapse [75]. Bazedoxifene therapy for up to 7 years was not associated with endometrial hyperplasia or changes in endometrial thickening, but was associated with a significant decreased risk of endometrial carcinoma (0.1 vs. 0.4 %; P = 0.020) [76].

Signals of a possible increased risk of prostate cancer in a clinical trial evaluating an oral preparation of salmon calcitonin in patients with osteoarthritis prompted a thorough review of all clinical trials of nasal and oral salmon calcitonin. A review by the EMA Committee for Medicinal Products for Human Use identified an increased risk of cancer (2.4 % with nasal administration) [78]. A meta-analysis of 17 randomized, controlled clinical trials with nasal spray salmon calcitonin, performed by the FDA, reported that the overall incidence of malignancies reported was higher among calcitonin salmon-treated patients (4.2 %) compared with placebo-treated patients (2.9 %) (OR 1.4; 95 % CI 1.1–1.7) [79]. Similar results were observed in a meta-analysis by independent authors [80]. No specific type of cancer was associated with calcitonin use. The authors of each report conceded that most of the clinical trials were poorly designed to assess new cancer cases but that the weak cancer signal could not be ignored. On the basis of this possible association with cancer risk, coupled with weak evidence of efficacy, European and Canadian regulatory authorities withdrew approval for nasal calcitonin as a treatment for osteoporosis [78]. In the United States, a FDA Advisory Committee recommended that the drug be withdrawn from the market [81]. The FDA decided instead to simply add a caution about the possible association of cancer risk [82]. Salmon calcitonin therapy has been associated with a few cases of serious allergic-type reactions including bronchospasm and anaphylactic shock, including rare reports of death [83].

The overall safety profile of denosumab has consistently been excellent in multiple clinical trials. The frequencies of adverse events, serious adverse events, and the rate of patients discontinuing therapy because of an adverse event have been similar between treatment and control groups. Among the 7805 women in the FREEDOM study , 90 in the placebo and 70 in the treatment group died (p = 0.08) [2]. Other than rare cases of ONJ and femoral shaft fracture with atypical features, no additional safety issues were clearly noted in the FREEDOM extension study [86]. A few cases of atypical fracture occurring with the clinical use of denosumab have been reported, but to date, all those patients had been pretreated with bisphosphonates [87]. Too few cases of atypical fracture have occurred to assess whether the risk, if present, is related to duration of therapy.

In the 3 years of FREEDOM, skin rash or eczema occurred more frequently with denosumab therapy than placebo [2]. The incidence of serious adverse events related to cellulitis was greater with denosumab (1.2 %) compared to placebo (one case among 3850 patients). These skin events were unrelated to the time or site of the denosumab injection. The cases of cellulitis responded to standard antimicrobial therapy. During the first 3 years of the FREEDOM extension, the incidence of skin rashes or serious adverse events related to cellulitis was very low in patients who continued taking denosumab for 6 years [86]. Importantly, increased risks of infections, eczema, or serious adverse events related to cellulitis were not observed in patients who had received placebo during the first 3 years and then received denosumab for the next 3 years.

RANK ligand is expressed in dendritic cells and some T cells, raising the possibility that immune dysfunction could result from inhibition of RANK ligand with denosumab [88]. The numeric incidence of infections and neoplasms was slightly greater among the denosumab in patients in FREEDOM compared to placebo. However, due to the manner in which adverse events are coded in clinical trials, many of these “infections” were inflammatory conditions such as diverticulitis or labyrinthitis rather than diseases caused by microorganisms [89]. There was no evidence of increased risk of opportunistic infections or immune-related neoplasms. No evidence of a progressive increase risk of either infection or malignancy was observed in the FREEDOM extension study. Rare, isolated cases of anaphylaxis associated with denosumab therapy have been reported [90, 91].

Denosumab, like other potent anti-remodeling agents, has potential to induce hypocalcemia. This complication of therapy was not observed in the FREEDOM study or numerous other phase 2 and phase 3 studies. However, all patients in those studies received calcium and vitamin D, and patients with vitamin D deficiency were excluded. Isolated case reports of hypocalcemia, sometimes severe and prolonged, have been described [92–94]. The risk of hypocalcemia may be greater in patients with significantly impaired renal function or with other metabolic bone diseases, particularly those associated with hypoparathyroidism, osteomalacia, and impaired bone mineralization [95]. Vitamin D deficiency should be corrected prior to starting therapy, and treatment is contraindicated in patients with hypocalcemia.

Few safety issues other than hypercalcemia and hypercalciuria were noted during the pivotal clinical trials. Muscle cramps occurred more frequently with teriparatide therapy, sometimes resulting in discontinuation of treatment. Regulatory restriction limits the duration of treatment to 18–24 months, the median length of treatment in pivotal trials, although effects on bone metabolism and fracture risk appear to persist for at least 3 years of teriparatide treatment in patients receiving glucocorticoids [99].

Mild hypercalcemia, usually transient, occurred in 11 % and 28 % of patients who received teriparatide 20 μg or 40 μg daily, respectively, compared to 2 % in the placebo group [96]. With the 20 μg daily dose, 95 % of patients with hypercalcemia had values less than 11.2 mg/dl (2.80 mmol/l). When measured at multiple times over 12 months, teriparatide 20 μg daily was associated with an increase in 24-h urinary calcium excretion from baseline by up to 32 mg/day compared with placebo at the same time point (P < 0.05) [100]. Hypercalcemia or hypercalciuria infrequently caused discontinuation of treatment. The higher incidence of hypercalcemia in the 40 μg group was the one reason that the 20 μg dose of teriparatide received regulatory approval. This agent is to be used with caution in patients with hypercalcemia or history of renal stones [101].

In rats receiving large doses of teriparatide for most of their lifetime, a dose-dependent risk of osteosarcoma was observed [102]. This resulted in regulatory warnings about the potential for osteosarcoma with teriparatide. Therapy is contraindicated in patients at risk for osteosarcoma including patients with Paget’s disease of bone, unexplained elevations of serum alkaline phosphatase, children and adolescents with open epiphyses, and patients with a history of skeletal radiation [101]. Isolated cases of osteosarcoma have been reported in patients with exposure to teriparatide, but the number of reported cases is not greater than anticipated in untreated older adults [103–105]. The Osteosarcoma Surveillance Study, an ongoing 15-year post-marketing surveillance study initiated in 2003, evaluates the potential association between teriparatide and development of osteosarcoma. After 7 years of surveillance, no signal of a causal association between teriparatide treatment and osteosarcoma in humans has been observed [106]. Because of the possibility of activating skeletal metastases, teriparatide should not be used in patients with malignancies involving or potentially involving the skeleton.

In the two major clinical trials, diarrhea, nausea, headache, and skin rashes occurred more often with strontium ranelate than with placebo, most commonly in the first 3 months of treatment. Elevation of serum creatine kinase occurred more frequently with therapy than with placebo [111]. Patients with marked renal impairment of renal function were excluded from the clinical trials, and the drug is not recommended in patients with calculated creatinine clearance of <30 ml/min [112].

An increased risk of venous thrombotic events was noted after 5 years of strontium ranelate therapy [112]. Deep vein thrombosis or pulmonary embolism occurred in 2.1 % of the placebo group and in 2.7 % of treated patients (OR 1.30, 95 % CI 0.90–1.88). Reviews of cohorts in the United Kingdom have not confirmed an increased incidence of venous thromboembolic events (VTE) [113–115]. Therapy is contraindicated in patients with current or previous VTE and those with temporary or permanent immobilization or prolonged bed rest.