Background

Bisphosphonates have proven to avoid osteoporotic fractures and are widely used to prevent these fractures particularly in postmenopausal women [1]. Prevalence of bisphosphonate use has considerably increased in the last years, due in part to demographic trends in older female population and to concerns of Hormone Replacement Therapy safety [2]. However, concerns have also been raised about serious adverse effects of this medication; these include osteonecrosis of the jaw, atrial fibrillation, atypical fractures [3] and, recently, conflicting evidence regarding risk of esophageal cancer [4]. Besides, a link between taking bisphosphonates and a raised risk of severe musculoskeletal pain has been established [5, 6]; indeed, prescribing information for all bisphosphonates already includes warnings of this particular risk [7, 8]. Additionally, cases of synovitis, a well-recognized cause of carpal tunnel syndrome (CTS) (acting through increased local pressure), have occasionally been reported [9–11].

CTS is the commonest peripheral nerve condition and has considerable employment and healthcare costs [12]; it is caused by elevated pressure in the carpal tunnel, and this in turn produces compression of the median nerve, resulting in impaired nerve conduction, paresthesia and pain [13]. The role of occupation has been the focus of much attention as a possible risk factor [14]. In a study carried out with a twin registry in the UK, a genetic predisposition was found to be the single strongest factor in predicting CTS [15]; BMI increase has also been consistently identified as a risk factor [16, 17]. Other medications such as lithium [18], oral contraceptives [19], exemestene [20]—an aromatase inhibitor used to treat breast cancer—and dabigatran [21] have occasionally been involved in CTS. The aim of this study was to explore whether bisphosphonates may be associated with carpal tunnel syndrome.

Methods

Results

Out of a sample of 59,475 women older than 51 years, 19,825 were treated with bisphosphonates during the period studied; this accounts for an exposed cohort of 51,245 person-years of follow-up. Distribution of prognostic factors between exposed and unexposed women after matching is presented in Table 1. No differences in age distribution or mean follow-up time were observed between the two groups in comparison. Overall, there were 572 women diagnosed with carpal tunnel syndrome, 242 (1.2%) in the group exposed to bisphosphonates, and 330 (0.8%) in the unexposed. An adjusted hazard ratio of developing carpal tunnel syndrome of 1.38 (95%CI, 1.15–1.64) was found for women exposed to bisphosphonates (Table 2); similar results were found when applying other methods to deal with missing values—complete subject analysis and missing data as a category. No significant changes in the hazard ratios were found when considering the levels of bisphosphonates exposure (Table 2). When considering those CTS codes referred to surgical procedures, the numbers of cases of CTS were 85 (0.4%) in the group exposed to bisphosphonates, and 115 (0.3%) in the unexposed; the crude and adjusted hazard ratios were, 1.49 (95%CI, 1.12–1.97) and 1.48 (95%CI, 1.09–2.00), respectively. Women exposed to aminobisphosphonates did not present a higher risk of CTS; adjusted hazard ratios of aminobisphosphonate-treated women (n = 15639) versus unexposed women and non-aminobisphosphonate-treated women (n = 2281) versus unexposed women were 1.36 (95%CI, 1.11–1.66) and 1.37 (95%CI, 0.95–1.96), respectively. For women exposed to alendronate (n = 11563), the most widely used bisphosphonate, the adjusted hazard ratio of CTS was 1.34 (95%CI, 1.08–1.65).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristics.

Distribution according to prognostic factors.

https://doi.org/10.1371/journal.pone.0146772.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Hazard of carpal tunnel syndrome associated with the intake of bisphosphonates.

https://doi.org/10.1371/journal.pone.0146772.t002

A higher BMI was independently associated with a higher risk of CTS; adjusted hazard ratios for BMI 25.0–29.9 versus BMI<25 and BMI >30 versus BMI<25 were 1.28 (95%CI, 1.06–1.59) and 1.80 (95%CI, 1.46–2.23), respectively. No statistically significant relationship was found between diabetes, hypothyroidism or rheumatoid arthritis and risk of CTS after controlling for other risk factors.

Discussion

In this cohort study with prospectively recorded information on prescribing of bisphosphonates, we found a statistically significant increased risk of CTS in women over 51 years of age treated with oral bisphosphonates. The risk was not related to the level of drug exposure, nor has it varied with the class of bisphosphonate used, whether amino- or non-aminobisphosphonate or even specific bisphosphonates such as alendronate. Our study was able to clearly confirm higher BMI as an independent risk factor for CTS; a higher BMI has been similarly identified in most of the observational studies upon CTS [16, 17]. However, though women with the highest BMIs are more likely to develop CTS, they are less prone to develop osteoporosis and subsequently less prone also to receive a bisphosphonate prescription; in fact, in our cohort, a lower percentage of women with the highest BMIs received bisphosphonate prescriptions (46% and 34% of women having BMI 25.0–29.9 and BMI >30 received bisphosphonate prescriptions, respectively, versus 68% for those having BMI <25). In our analysis comprising women whose median age was 72 years-old we did not observe diabetes, hypothyroidism and rheumatoid arthritis as significant risk factors for CTS; this is consistent with what has recently been observed in a large case-control study [26], aging appears to reduce the relative impact of the diseases commonly associated with CTS as the possible risk factors.

An alert highlighting the possibility of severe and sometimes incapacitating bone, joint, and/or musculoskeletal pain has been issued by the FDA [27]; indeed, prescription information for all bisphosphonates includes warnings of this particular risk [7, 8]. This severe musculoskeletal pain may occur within days, months or years after starting bisphosphonates. A series of 117 cases of severe musculoskeletal pain developing in adults on bisphosphonates has been published [5]. Moreover, cases of synovitis confirmed by positive re-challenge have also been reported [9–11] and may occur at any point after starting bisphosphonate therapy, similarly to the occurrence of the cases of CTS we identified.

Arthritis—one of the leading causes of CTS [10]—and its correlate arthralgia is one of the most reported adverse effects for bisphosphonates. Two facts would be able to explain biological plausibility of bisphosphonate-induced arthritis and synovitis. First, the appearance of inflammatory cytokines, in particular TNFα and IL6, has been observed after bisphosphonate treatment in vitro and in patients [28–31]; thus, for synovitis to occur, synovial macrophages would release those cytokines when bisphosphonates are present; the way this process is triggered remains unclear. Second, the skeletal half-life of various bisphosphonates is sometimes more than 10 years [32]; once bisphosphonates are buried in the skeleton, they will be released only when the bone is destroyed in the course of turnover. Being an inflammatory process which can be triggered by these substances in any moment during long periods would explain the fact that this reaction is not related to time or dose.

The strengths of the present study are its large sample size, the long period of follow-up, and the use of recorded prescription data rather than self-reported drug use, which may misclassify exposure.

Since bisphosphonates cannot be obtained without prescription in the United Kingdom, underestimation of its usage would seem unlikely; however, a limitation of the study is that drug adherence is unknown. In fact, since exposure does rely on prescriptions issued, overestimation of usage is possible, as compliance with bisphosphonate prescribing is known to be suboptimal [33]); if so, the real CTS risk associated with bisphosphonates would be higher. Additionally, an interval between prescription and intake might exist; however this information is unknown for the two compared groups. Also, we relied on relevant diagnostic codes from patients’ clinical files and CTS is a specialist diagnosis. However, as codes have a certain degree of unreliability, some misclassification might be inevitable; therefore, we considered those codes involving CTS surgical procedures which further supported the accuracy of the diagnoses—most surgical procedures would also require electrophysiological testing, although this is not routinely recorded in the notes—, finding no significant changes in our previous estimates. Furthermore, the treatment of missing data for BMI, the only important variable with a considerable number of missing data, was handled using the current recommended approaches to avoid the biases that occur when omitting individuals with incomplete data. We were not able to adjust for vocational activities as data was incomplete similarly data for chronic kidney disease was sparsely recorded; although we have adjusted for the number of visits, it is still possible, but unlikely, as a surveillance bias.

In summary, we have identified an increased risk of CTS associated with bisphosphonates; it is a novel finding with potential clinical impact. As increasing numbers of individuals, particularly postmenopausal women, are exposed to this medication, a large number of cases of this invalidating and costly condition might appear in excess. If confirmed, doctors should be aware to avoid prescribing this medication in those women with other risk factors for CTS such as elevated BMI or family history of the disease; regulatory authorities, in turn, should consider appropriate further safety advice regarding CTS risks with bisphosphonates.