Abbreviations
AACE
American Association of Clinical Endocrinologists
ALN
alendronate
ALP
alkaline phosphatase
ASBMR
American Society for Bone and Mineral Research
AFF
atypical femoral fractures
BMD
bone mineral density
BP
bisphosphonate
BTM
bone turnover marker
CTX
C-telopeptide of Type 1 Collagen cross-links
CPG
clinical practice guideline
Dmab
denosumab
DXA
dual-energy X-ray absorptiometry
FNIH
Foundation for the National Institutes of Health
FLEX
Fracture Intervention Trial Long-term Extension
FDA
Food and Drug Administration
FRAX
fracture risk assessment tool
HT
hormone therapy
HRT
hormone replacement therapy
IBAN
ibandronate
NOGG
National Osteoporosis Guideline Group
NTX
urinary N-telopeptide cross-links
ONJ
osteonecrosis of the jaw
PINP
N-terminal propeptide of type I procollagen
PTH
parathyroid hormone
RIS
risedronate
RRR
relative risk reduction
SERM
selective estrogen receptor modulator
TPTD
teriparatide
ZOL
zoledronic acid
79.1
Introduction: osteoporosis as a chronic disease
Osteoporosis is a chronic disorder and a major health problem, with an estimated 2 million osteoporotic fractures every year in the United States, and over 3.5 million in Europe. As life expectancy increases in many parts of the world, the annual incidence of osteoporotic fracture is destined to rise, and more people will be potential candidates for treatment.
Peak bone mass is attained in the third decade of life, and thereafter bone mass decreases. In women the decrease is accelerated by the loss of estrogen at the time of menopause. It is clear, however, that bone mass does not fully explain the age-related increase in fracture risk. If two people, ages 55 and 75 years, have the same bone mass as reflected by dual-energy X-ray absorptiometry (DXA), the older patient will have a much higher fracture risk than the younger ones. This difference is ascribed to decrease in bone quality, frailty, and tendency to fall [as best expressed in the age-dependent hip fracture incidence in the fracture risk assessment tool (FRAX) calculator] but yet still difficult to directly measure. Nonetheless, it can be stated that aging is a major risk factor for fracture and getting older increases fracture risk for the population overall. For those individuals with low bone density by DXA and other risk factors, as covered in other chapters, fracture risk continues to escalate.
Currently, no treatment cures osteoporosis. All existing treatments lower fracture risk to some extent, but none can eliminate it. Hence, osteoporosis can be classified as a chronic treatable but not curable disease. Management should be aimed at lowering fracture risk and containing related morbidities and costs to individuals and society as much as possible. Osteoporosis may require management for 20 or 30 years or more, and in most settings, bisphosphonates (BPs) are the first-line osteoporosis treatment class. Yet the longest placebo-controlled osteoporosis pharmacologic study lasted 10 years and was too small to adequately assess positive and negative outcomes, as well as predictors of such outcomes. Therefore the clinician must base treatment choices on relatively limited data. In this chapter, we will discuss how long-term treatment with BPs can decrease fracture risk, conduct a risk–benefit assessment of the positive impact of BPs and their potential side effects, and provide a proposed approach to the chronic treatment of osteoporosis with BPs.
79.2
Long-term bisphosphonate studies
There are many reasons why the BP class of medication is the single most commonly used category of drugs for the treatment of osteoporosis in adults. Modern treatment of osteoporosis began in the mid-1990s with the approval of alendronate (ALN), the first nitrogen-containing BP approved by regulatory agencies for human osteoporosis. Details of the studies assessing the efficacy and safety of ALN, risedronate (RIS), ibandronate (IBAN), and zoledronic acid (ZOL) are beyond the scope of this chapter. In general, the data show that all US Food and Drug Administration (FDA)–approved BPs lower the risk of vertebral fracture by about half . In addition, ALN, RIS, and ZOL reduce hip fracture by about 30%–40%. The efficacy of nonvertebral fracture risk reduction is more modest. The registration trials for ALN, the earliest approved nitrogen-containing BP, did not provide evidence of any major side effects. Only later, when hundreds of thousands of patients were treated with BPs, did important risks such as osteonecrosis of the jaw (ONJ) and atypical femoral fractures (AFFs) become recognized. Longer term treatment and surveillance are necessary because no treatment cures osteoporosis. With the use of BPs for many years, and with long treatment duration, more reports of ONJ and AFF became apparent, leading to dramatic decreases in prescriptions for these drugs . Indeed, it has been reported that 1 year after starting an oral BP, only half or fewer of treated patients remain adherent to therapy .
Thus the clinician and patient face a dilemma. Osteoporosis is a chronic disease with worsening fracture risk as the patient ages. Chronic treatment is needed, but BPs, the most commonly prescribed osteoporosis medications, are associated with risks of ONJ and AFF. Although such adverse events are rare and by far outweighed by substantial benefits in terms of fracture risk reduction in high-risk patients, they are clearly problematic in lower risk individuals. There may be other serious side effects. How does the clinician decide to prescribe BPs, to continue or interrupt therapy, and engage the patient in shared decision-making? Risk stratification, absolute risk prediction, and a risk–benefit assessment are crucial to such decision-making. To try to answer these questions, there are only two long-term randomized controlled trials on which some recommendations have been made. These two trials, a 10-year trial with ALN and a 9-year trial with ZOL, will be reviewed in detail.
79.2.1
Alendronate (FIT/FLEX)
The first study of long-term BP therapy for osteoporosis, the FLEX trial (Fracture Intervention Trial Long-term Extension), was reported in 2006 by Black et al. . Over 3000 women who had participated in the FIT trial of ALN versus placebo and had received active drug were assessed for potential inclusion in the extension trial. With an open-label year or more, subjects eligible for FLEX had been treated for 4–5 years. Eventually, 1195 were screened for inclusion in FLEX and 1099 were randomized to ALN 5 mg daily, ALN 10 mg daily, or placebo. All subjects were offered a supplement containing 500 mg of calcium and 250 U of Vitamin D. The FLEX study lasted 5 years with more than 95% included in the primary analysis. The randomization and block stratification were successful because the only baseline variable different among the three groups was the slightly older age of the placebo group (73.7 years) than the two ALN groups (72.7 and 72.9 years). The two ALN groups were pooled for analysis. Bone density results differed by region of interest. In the spine, bone density increased in subjects on ALN, while it did not change in the placebo group. Conversely, hip bone density was maintained in subjects taking ALN, but it declined 2.36% in the placebo subjects . Of the 437 analyzed placebo subjects, 23 suffered a clinical vertebral fracture, compared to only 16 of 662 subjects assigned to ALN. There wasno difference in the incidence of morphometric spine fractures or other clinical fractures, and no imbalance in serious adverse events between the groups. The conclusion from the FLEX trial was that 10 years of ALN led to fewer clinical vertebral fractures than 5 years of ALN followed by 5 years of placebo (relative risk 0.45, confidence interval 0.24–0.85) .
79.2.2
Zoledronic acid (HORIZON)
The other long-term treatment study was the HORIZON extension trial in which women who had received three annual infusions of ZOL were rerandomized to receive either 3 more years of ZOL or three placebo infusions . Of the 2629 women who received 3 years of active drug, 1233 underwent rerandomization. Of the 616 allocated to receive three more ZOL infusions, all but three women received at least one additional infusion. Three annual doses were received by 451 women, but 44 did not complete the follow-up. The final bone mineral density (BMD) analysis included 451 women who received at least one additional dose, and the final X-ray analysis included 469 who received at least one additional dose. For the group rerandomized to placebo infusions, the corresponding numbers available for DXA and X-ray were 470 and 486, respectively. Bone density in the hip tended to decrease in the placebo group compared to the group that received three more ZOL acid infusions, but the difference between them was just over 1%. Spine bone density increased in both placebo and active drug groups but was greater by 2% in the group that received the second 3-year set of ZOL infusions. Interestingly, at year 6, there was very little difference in a bone resorption marker, C-telopeptide of type 1 collagen cross-links (CTX), between the two groups, suggesting that the first three infusions of ZOL had a lasting impact on bone resorption. While there were no differences in clinical fractures between the two groups, there were fewer morphometric vertebral fractures in the women who received six infusions of ZOL compared to those who only received three .
In addition to the first extension of HORIZON, 190 women who had received six annual infusions of ZOL met inclusion criteria and agreed to be randomized once more to either the more active drug or 3 placebo infusions . About two-thirds had bone density measurements at the end of the trial. Interestingly, hip bone density tended to decrease slightly in both groups, with no significant difference between them. Again CTX was also the same in the two groups, again suggesting a long-term effect of prior ZOL infusions. There was a small imbalance in cardiac arrhythmias in those women who received nine ZOL infusions compared to six, although there was no difference in arrhythmias as serious adverse events. The study was too small to observe any differences in fracture incidence .
In a recent post hoc analysis, bone density and bone formation markers were compared at 3 years after BP discontinuation in FLEX and the HORIZON Extension Trial . The purpose of this study was to determine the impact of a drug holiday, after the recommendation that one be considered after 5 years of oral treatment, similar to the FLEX study or 3 years of intravenous treatment, similar to the HORIZON trial. At the start of the extension studies, there were some differences in-between the women in FLEX and the women in HORIZON. The subjects in the former group were significantly younger (73.7 vs 75.5 years) and had fewer prevalent vertebral fractures (34.3% vs 63.2%) and higher hip bone density ( T -score −1.8 vs −2). Three years after the start of the extension period, more subjects previously on ALN had a significant decrease in total hip bone density compared to those previously treated with ZOL (25.2% vs 18.7%, P <.01). This may again suggest a long-lasting effect of ZOL .
Other studies confirm the extended impact of ZOL infusions. In 2009 McClung et al. randomized 581 postmenopausal women (average age about 60 years) to two annual infusions of ZOL, one infusion of ZOL at baseline and a placebo infusion at 1 year, or two annual placebo infusions. At 2 years, total hip had increased 2.91% in the first group and 2.28% in the second but decreased 1.45% in the placebo group. The pattern was similar in the femoral neck. Serum CTX declined markedly in the first group with a slow rise toward the end of month 12. With the second ZOL infusion the CTX decreased again. In contrast, a single ZOL infusion at baseline was followed by a gradual increase of CTX, although it never reached the baseline level. The placebo group had no change in CTX over the 2-year study. Greenspan et al. reported on the 2-year impact of a single ZOL infusion versus placebo in older women (average age about 85 years) . At 2 years, total hip bone density increased by 2.6% in the active drug group and decreased by 1.5% in the placebo women. However, in these older women, improvement in femoral BMD neck at 1 year was not sustained to 2 years, although the absolute value was still higher than the placebo group. In older women the ZOL-induced drop in CTX was sustained for 2 years.
Grey et al. administered one infusion of varying doses of ZOL acid to women in their mid-sixties . Comparing those that received the standard 5 mg dose to placebo, total hip BMD was still above baseline 5 years after the infusion and 5.4% higher than that of the placebo subjects. Also, at 5 years after the single infusion, CTX was 27% lower in the active drug group compared to placebo. Recently, Reid et al. reported the impact of four infusions of ZOL (5 mg each) over 6 years compared to placebo in osteopenic women aged about 71 years . Total hip BMD was almost 4% higher at 6 years in the active drug group, whereas it had decreased almost 4% in the placebo group. There were 37% fewer fragility fractures in the ZOL group, compared to placebo. The conclusion from the cited studies is that the impact of ZOL on surrogates for fracture, BMD, and bone turnover markers (BTMs) lasts longer than 1 year. Thus a patient treated with three annual infusions of ZOL has evidence of continued effect beyond 3 years and may be eligible for an interruption in treatment. It has also been suggested that the interval between ZOL infusions could be lengthened such that a patient would have three infusions over 5 years , similar to the four infusions in 6 years as reported by Reid et al. . Having all BP-treated patients take 5 years of treatment—either oral for 5 years or three infusions spread over 5 years—simplifies management for the busy clinician, but studies of the efficacy of the latter regimen on fracture risk reduction are lacking.
79.2.3
Risedronate and ibandronate
There is much less information about long-term treatment with RIS or IBAN. Eastell et al. reported on subjects who had 5 years of blinded RIS treatment (5 mg/day) followed by a 2-year open-label continuation of the same dosage . The subjects then stopped RIS and were assessed at 6 and 12 months after discontinuation. Urinary N-telopeptide (NTX) rose toward baseline at the two time points. BMD was measured at the 1-year discontinuation visit. While lumbar spine and femoral neck BMD were maintained, total hip BMD decreased back to the about the original (year 0) level. In an earlier study, subjects discontinuing RIS after 3 years of treatment remained at lower risk for morphometric vertebral fracture compared to subjects who had been on placebo for 3 years followed by the 1-year extension . In an extension trial , women on various regimens of IBAN for 2 years were randomized to monthly IBAN in doses of 100 or 150 mg. Over the next 3 years, spine BMD rose modestly in both groups (no difference) and total hip was maintained. There was no placebo group; thus information on treatment interruption is lacking.
79.2.4
Observational data
Observational studies have provided some information on long-term treatment and interruption of treatment. However, such studies have serious limitations, including confounding by indication and by low persistence with BP treatment. In a recent systematic review, persistence with BP treatment ranged from 17.7% to 74.8% at 1 year and 12.9% to 72.0% at 2 years . At 3 years the medication possession ratio varied from 27.2% to 46%. Tracking of BP use is often complicated by patients stopping and restarting treatment. For example, Balasubramanian et al. used data from the Truven commercial insurance database in the United States . They found that by 2 years, 70% of women had discontinued oral BP treatment. Almost half of women who discontinued treatment restarted it; those women who were hospitalized or were older were less likely to reinitiate BP therapy. The two largest observational studies are from the population-based Kaiser Permanente and Medicare databases. The Kaiser report included over 39,000 women, age 45 and above, who took BPs for more than 3 years compared to women who went on a drug holiday . Fracture risk was lower in the drug holiday group suggesting that in that group, BP discontinuation was due to a lower risk profile. Conversely, a Medicare database report included over 160,000 highly compliant older women, and showed a RR of hip fracture of 1.22 (1.11–1.34) at a median of 2.7 years post drug interruption, a risk that increased with the duration of the drug holiday. In countries with medication registries, it may be easier to determine real-world effectiveness of treatment. Strom et al. reported that the longer adults stayed on BP treatment, the less likely they would be hospitalized for a fracture . These findings are compatible with the conclusions of the IOF Epidemiology/Quality of Life Working Group that discontinuation of BPs leads to increased risk of fracture .
79.2.5
Side effects of bisphosphonates
A full discussion of BP’s side effects is beyond the scope of this chapter. Two side effects, however, ONJ and AFF, play a major role in long-term BP treatment because the fear of such side effects and their sensational reporting in the general media have had a profound impact on osteoporosis evaluation, diagnosis, treatment, and adherence, despite their low incidence. ONJ is defined as at least 8 weeks of exposed bone in the maxilla or mandible, despite appropriate therapy. Most definitions also require that there has been no radiation to the jaw or metastatic disease in the area . BP-associated ONJ was described as early as 2003 in patients with metastatic disease receiving high and repeated doses of the intravenous BPs, pamidronate, or ZOL. Although patients may be asymptomatic, many have pain, swelling, and in severe cases a fistula. Risk factors include smoking, poor oral hygiene, diabetes mellitus, glucocorticoid therapy, and invasive dental procedures. Current estimates for BP-associated ONJ in patients on BP for osteoporosis range from 1/10,000 to 1/100,000 . A recent review from Denmark found the incidence of surgery for ONJ to be about 2.53 per 10,000 patient-years in patients on ALN . There was an increased risk for recent, long-term, and adherent users of this most commonly used BP for osteoporosis. The increased risk with long-term use is consistent with findings from an early series . In another recent study the number of women in the United Kingdom who were admitted to hospital because of ONJ was tracked over 8.2 years . In women without cancer the incidence of hospitalization for ONJ was 1.38 per 10,000 patient-years in oral BP users compared to 0.18 per 10,000 patient-years in never users. In patients with osteoporosis the course is usually self-limited and can be treated conservatively . These studies are good examples of a serious but unusual side effect that, nevertheless, has led to fewer patients initiating or adhering to BP treatment for osteoporosis. In the 2015 International Task Force report on ONJ, Khan et al. cited the highest incidence of ONJ in the oncology patient population (1%–15%), where high doses of these medications are used at frequent intervals, in contrast to an incidence of 0.001%–0.01% in the osteoporosis patient population, an incidence that is marginally higher that in the general population (<0.001%) .
Even more worrisome in the general population is the risk of AFF. The idea that a drug taken to reduce fracture risk might cause a particularly serious fracture was reported widely in popular media . AFFs, as defined by the American Society for Bone and Mineral Research (ASBMR) Task Force, are located on the femur from just distal to the lesser trochanter to just proximal to the supracondylar flare and must have four of five major characteristics: minimal trauma at most, a start at the lateral cortex, and be mostly transverse, if complete will have a medial spike (incomplete will be in lateral cortex only), no or minimal comminution, and the lateral cortex will have a localized reaction that shows “beaking” or flaring . While such fractures have been reported in patients never exposed to BP (or other antiresorptive drugs) and in patients with genetic bone disorders not treated with BPs , there is no doubt about an association with BP treatment, with increased risk with longer duration and a decrease in risk with treatment cessation. An observational study from Italy provides perspective on the incidence of AFF versus typical osteoporotic fractures in a 7-year analysis (almost 1.4 million patient-years) of a major hospital . During this time, there were 4003 fractures of the femur, of which 3335 were low trauma and in the femoral neck or trochanteric region. There were 308 low trauma fractures that were subtrochanteric or in the femoral shaft. Of these, only 22 were atypical (13 patients on BPs). Of the non-AFF in the subtrochanteric or shaft area, 20 of 286 patients were on BPs. Other epidemiologic studies have confirmed the fact that osteoporotic fractures of the proximal femur are much more common—and can be decreased by BPs—than AFF caused by BPs. A recent review made several estimates of benefits versus harms of AFFs . In a 3-year period, they estimated that for every 10,000 patients treated, 1000 fractures would be prevented and just under 1 atypical fracture would be caused. However, treatment is almost always for longer than 3 years, and there is good evidence that duration of BP treatment increases the risk of AFF. In the widely cited Kaiser Permanente Study, the rate of AFF rose from 22 per 100,000 patient-years for patients on BPs from 4 to 5.9 years to 34.5 per 100,000 patient-years in those who took BPs for 10 or more years . Other studies have confirmed the increased risk of AFF with time on BP treatment . The impact of duration has led to suggestions that drug holiday or interruption of treatment might have a beneficial effect (see next). It has been difficult to determine which BP-treated patients are at highest risk for AFF, in addition to those on treatment for extended periods of time. Potential risk factors for AFF include the use of chronic glucocorticoids, proton pump inhibitors, and diabetes mellitus . In the United States, people of East Asian ancestry have been found to be at higher risk . Bowing of the femur, a shorter angle between the femoral neck and shaft, and varus alignment of the lower extremity suggest that the geometry of the pelvis and femur may increase the amount of tension on the lateral cortex of the femur. In practice, thigh or groin pain and possibly single energy images of the femur at the time of DXA testing may provide evidence of impending fracture . The 2019 ISCD official positions for adults recommend the use of femur DXA images to detect abnormalities in the spectrum of AFF and scanning methods that generate bilateral full-length femur images .
Regardless of the low incidence of AFF compared to typical proximal femur fractures, concern about AFF has been trumpeted by news media, leading to patients stopping or never starting BP therapy. Not on the news have been studies suggesting that BPs have a salutary effect on mortality, including a recent large well-designed prospective cohort study .
79.3
Bisphosphonate interruption (holiday)
79.3.1
Pharmacokinetics of bisphosphonates
A comprehensive review of the pharmacokinetics and pharmacodynamics of ALN, RIS, IBAN, and ZOL is well beyond the scope of this chapter. In addition, there remain major questions to be answered. From reviews of this subject , it can be concluded that oral BPs are poorly absorbed in the gastrointestinal tract, but both oral and intravenous BPs are mostly incorporated into bone. The retention time is measured in months to years, and the impact on BTMs is also measured in months to years. A study of urine measurements of ALN and RIS in women who stopped treatment is emblematic of the long terminal half-life . Postmenopausal women who took ALN (mean 51 months) or RIS (53 months) were assessed after they had discontinued treatment for 13–14 months. Using an HPLC method, the authors detected urinary ALN in 41% of the women who had taken ALN, but no RIS was found in the urine of previous RIS users. These data can inform the potential impact of discontinuation of osteoporosis treatment.
79.3.2
Placebo groups from FLEX and HORIZON
Review of the placebo subjects in the two main long-term osteoporosis trials provides prospective blinded data on the impact of BP discontinuation . As described earlier, in the FLEX trial, subjects previously on ALN but rerandomized to placebo had no change in spine BMD over the 5-year extension, but the total hip BMD decreased 2.8% . In the HORIZON trial extension after 3 years of annual ZOL infusions, spine BMD rose similarly in those women receiving three more annual active drug infusions or three placebo infusions . In the hip, there was little change in the 3-year extension. In FLEX, placebo women had more clinical vertebral fractures than those continuing ALN, whereas in the HORIZON extension, placebo women had more morphometric vertebral fractures. In neither study were there enough nonvertebral fractures to determine the impact of the discontinuation of active BP on fractures at these skeletal sites.
79.3.3
Observational studies of drug interruption (holiday)
Observational studies may have unmeasured confounders, but advantages of this study design include the potential for longer term follow-up and larger numbers of study subjects, including people who may not qualify for randomized controlled trials. In a small study of women and men who discontinued oral treatment for a mean of 6.5 years or intravenous BP for 3.6 years, there were minimal decreases in spine BMD at 24–30 months . However, total hip and femoral neck BMD decreased almost 3%. There was only one wrist fracture observed as well as 2 foot fractures, and interestingly a majority of the patients restarted osteoporosis treatment. Another observational study reported that in a follow-up period averaging about 2.5 years, the chance of fracture was 40% higher in women who discontinued BP treatment compared to those who remained on treatment . In a somewhat larger study, 62 of 401 women who stopped treatment for 5 years developed a fracture . Higher age and lower BMD were associated with greater chance of fracture. The largest observational study was from Kaiser Permanente, which is a closed system with excellent pharmacy records. Women who had taken BPs for at least 3 years were assessed . Those women who had maintained at least 50% adherence were considered persistent. Those who adhered less than 50% of the time or discontinued for a period of less than 1 year were considered nonpersistent. A drug holiday was considered if the patient discontinued treatment for more than 1 year (average length 3.1 years). The fracture rate (per 1000 person-years) during the study period was 32.8, 43, and 28.8 for the persistent, nonpersistent, and drug holiday groups, respectively. Thus this large observational study suggested that intermittent intake of BP led to most fractures. While the three groups were similar in many ways, it is quite possible that unmeasured factors were important in determining who fractured.
Three systematic reviews of osteoporosis treatment and discontinuation have been recently published. Fink et al. concluded that continuing treatment beyond the first conventional treatment period may lower vertebral fracture risk but may increase the risk of rare adverse effects . Based on the placebo groups of the long-term studies discussed earlier, Fink et al. also concluded that taking ALN for 10 compared to 5 years was associated with lower risk of clinical vertebral fracture and having six rather than three infusions of ZOL was associated with fewer morphometric vertebral fractures. Nayak and Greenspan concluded that women with low BMD at the end of the conventional initial treatment period were likely to benefit from continued treatment . Discontinuation of treatment should be considered for women with higher BMD (i.e., T -scores >−2.5) at the end of the initial treatment period. Conversely, based on their analysis of the existing evidence, Dennison et al. disagreed with the concept of drug holidays being always offered to patients on long-term BPs . Tailoring the decision for drug interruption is best based on each individual patient’s risk–benefit profile from long-term therapy.
After osteoporosis treatment discontinuation, BTMs are likely to change sooner than BMD. Naylor et al. studied women who had been treated with IBAN, ALN, or RIS for 2 years . Fifty-seven women with T -scores better than −2.5 completed the 2-year no-treatment extension, and 50 had BTM data measured. Two-thirds had changes in CTX beyond the least significant change, and almost the same number had levels above the reference range. Almost three-fourths had significant increases in P1NP. Interestingly, those women with the greatest change in BTMs had the largest decrease in total hip BMD off therapy (for CTX, r =−0.58 and for P1NP, r =−0.41). These data suggest that BTMs could serve as a tool for following patients after discontinuation. However, the FLEX trial showed that follow-up measurements of BTM were not associated with fracture incidence off therapy . Using a different BTM, urinary deoxypyridinoline cross-links (uDPD), in an observational study, Liel et al. found that most patients on ALN had a substantial rise in this marker after the withdrawal of treatment. In about one-quarter of these women, the uDPD increased to above the upper limit for premenopausal women. However, a clinically useful tool for determining when to restart osteoporosis treatment is not clear from these studies . Elevation of a BTM beyond the normal range might be a reasonable target, but it is likely that even within the normal range, BTM increases may signal return of increased fracture risk . In a recent study, Kim et al. compared changes in BMD and BTMs 3 years after switching to placebo from ALN in the FLEX study and 3 years after switching to placebo from ZOL in the HORIZON extension study. Subjects in the former study had greater loss of hip BMD and were more likely to have a significant rise in P1NP. The authors concluded that the offset of medication effect from ZOL was longer than from ALN. In clinical practice, BTM levels can be affected by many factors .
79.4
Guidance for patients on long-term bisphosphonate therapy
The worrisome reports of serious adverse events in patients on long-term BP in the literature several years after their drug release on the market culminated in three consecutive FDA safety-related announcements; the first on ONJ in 2005, the second on atrial fibrillation in 2007, and the third on AFF in 2010. Finally, in September 2011, after a hearing to review the long-term safety and efficacy of BPs, the FDA recommended that physicians reassess the indication for continued BP therapy beyond 3–5 years, but noted that drug discontinuation may not be advisable in high-risk patients. Currently, all FDA-approved BPs for the treatment of osteoporosis specify “Important Limitation of Use: The optimal duration of use has not been determined. All patients on BP therapy should have the need for continued therapy reevaluated on a periodic basis” .
The consecutive FDA warnings and intense media coverage about potential risks from prolonged BP therapy were paralleled by a series of spikes in internet search activity on ALN quantified by Google Trends . This was followed by a steep decline in BP use over the following years, including high-risk populations, such as patients with hip fractures, in the United States and worldwide . This development led several organizations to examine the evidence to provide the best possible guidance to their constituencies. A risk–benefit analysis, risk stratification, and the concept of drug “holiday” became a common approach incorporated by all organizations in their guidance on therapy continuation in patients on long-term BPs. Therapy interruption or suspension may be more appropriate terms, but holiday is the term most commonly used in the literature over the years.
79.4.1
American Society for Bone and Mineral Research 2016 Task Force Report: managing osteoporosis in patients on long-term bisphosphonate treatment
In 2013 the ASBMR convened a multidisciplinary Task Force on “Managing Osteoporosis Patients after Long-Term BP Treatment.” Osteoporosis experts in epidemiology, endocrinology, geriatrics, and drug surveillance served on the Task Force. The main charges were set by the ASBMR Professional Practice Committee (PPC), approved by Council, and subsequently slightly modified by Task Force members to address complementary themes. These were to
- •
“Provide guidance on duration of BP therapy in patients with postmenopausal osteoporosis, developing an algorithm that incorporates risk assessment (efficacy).
- •
Determine how potential harms may affect duration of therapy (safety), with a risk/benefit perspective.
- •
Discuss how the algorithm may apply to men and to glucocorticoid-induced osteoporosis.”
Three parallel systematic literature searches using Medline, EmBASE, Cochrane, and PubMed were implemented on the following: randomized controlled trials with long-term BPs, BPs and drug holidays, and BPs and guidelines. The FLEX and HORIZON extension studies provided the evidence to develop the algorithm. Other topics thoroughly reviewed by the Task Force members included BP benefits, harms, differential effects, their resolution upon drug discontinuation, monitoring on and off therapy, alternative therapeutic options, and patient adherence and preferences. Case studies were included to illustrate applicability of the proposed care pathways to challenging clinical scenarios. The Task Force work spanned 2 years, the ensuing report was reviewed and endorsed by the ASBMR PPC, and it underwent peer review by the JBMR editorial board prior to publication .
The derived management algorithm was based on post hoc analyses from FLEX and HORIZON extension, registration trials exclusively conducted in postmenopausal women and thus pertaining to this specific population. For postmenopausal women who have been on oral BP therapy for 5 years, or intravenous ZOL for 3 years, but less than 10 years, the Task Force suggested that oral BP therapy be continued for up to a total of 10 years for ALN, and 6 years of ZOL, in the highest risk stratum, namely, patients who had experienced a hip, spine, or multiple other osteoporotic fractures prior to therapy or those who experienced a major osteoporotic fracture (spine, hip, humerus, or forearm) on therapy. For the latter subgroup an evaluation for causes of secondary osteoporosis, appearance of new risk factors, vitamin D deficiency, and assessment of medication adherence was recommended. In addition, switching to alternative therapies was a possible option, but there were no adequate studies then to evaluate the efficacy of such an approach. The optimal length of therapy for the patient who suffers a fracture while on treatment also could not be defined, and clinical judgment was called on to address such cases. In individuals who have not experienced a fracture, other variables that can signal high fracture risk were recommended to inform decisions on therapy continuation. These were based on post hoc analyses from FLEX and HOROZON, and included low T -score (≤−2.5) and older age (>70–75 years). Other variables deemed indicative of high risk included a high fracture risk based on country-specific risk assessment tools, medication use (e.g., aromatase inhibitors, glucocorticoid therapy), or a new diagnosis of a disorder associated with secondary osteoporosis. If based on the above criteria the patient remained at high fracture risk, the Task Force suggested that BP treatment be continued for another 2–3 years, with periodic reassessment. Alternative antifracture therapy, with teriparatide (TPTD) or denosumab (Dmab) as first options, then raloxifene, could be considered for those patients remaining at high risk for fracture. For those women who are not considered to be at high fracture risk, a drug holiday of 2–3 years was to be considered, with monitoring and risk assessment every 2–3 years (possibly earlier for women on RIS). Tools to identify patients who will fracture off therapy and for monitoring are limited, and therefore the Task Force report suggests reinitiation of therapy, using BPs or alternative therapies, if a patient’s T -score drops below −2.5, or additional risks appear . Although the report did not support routine measurement of bone remodeling markers, it noted that some experts would resume therapy in patients when they exceed the lower half of the premenopausal range, off therapy . The report also examined data in men and subjects on glucocorticoids and concluded that the algorithm would probably be applicable to men and patients with glucocorticoid-induced osteoporosis, with some modifications, such as raising T -score thresholds for risk stratification in patients on steroids.
In its risk–benefit assessment the report compared the incidence of ONJ and AFF and that of typical osteoporotic fractures, as well as that of other important outcomes and serious events. BP therapy for up to 5 years will prevent approximately 175 hip fractures, 1470 vertebral fractures, and 945 wrist fractures (2590 total/100,000) for 16 AFFs/100,000 associated with treatment. This would equate into a total of 162 fractures of the spine, hip, or forearm prevented/AFF potentially caused. For longer treatment, there are insufficient data to derive similar estimates .
The Task Force report underscored several limitations to the evidence available and thus to its approach, in its abstract, main text, and conclusion. The algorithm reflected data from clinical trials in which the majority of subjects were Caucasian women (from Europe and America). It was based on limited evidence and applies only to vertebral fracture reduction, mostly in white postmenopausal women (from Europe and America), and not necessarily to other BPs. Furthermore, the reduction in vertebral fractures differed between FLEX and HORIZON. The reduction was in clinical vertebral fractures with 10 years of ALN (compared to 5 years of ALN and 5 years of placebo) and for morphometric vertebral fractures with 6 years ZOL (compared with 3 years ZOL and 3 years of placebo). Finally, the report outlined that country-specific thresholds and those for non-Caucasian women vary for initial treatment and therefore this may also affect thresholds for continuation or reinstitution of therapy. In view of all previously limitations the report therefore used the terms guidance rather than guideline, “suggested approach,” and verbs such as “may” and “would” rather than should, consistently in the document. The abstract reads “It is obvious that there is relatively little evidence on which the Task Force can base recommendations, and indeed we have presented management suggestions based on limited data and clinical experience.” A call for clinical judgment was made and for the ultimate decision to continue long-term BP therapy beyond 5 years to take into consideration previously limitations, patients’ values, and preferences, to enable an individualized approach. Its approach was incorporated in the algorithm proposed by the Endocrine Society in its clinical practice guideline (CPG) on “Pharmacologic Management of Osteoporosis in Postmenopausal Women.”
79.4.2
Endocrine Society 2019 Clinical Practice Guideline: pharmacological management of osteoporosis in postmenopausal women
In 2016 the Endocrine Society convened an international panel of experts with the main objective to formulate a CPG for the pharmacological management of osteoporosis in postmenopausal women . The committee included endocrinologists and osteoporosis experts from Canada, the United States, and Europe and a methodology expert. It commissioned two systematic reviews, one to quantify the evidence for the efficacy of various pharmacologic interventions to reduce vertebral, hip, and nonvertebral fractures and the other to assess patients’ preferences in a qualitative approach. The first systematic review and metaanalysis included 107 trials (193,987 postmenopausal women; mean age of 66 years; 55% Caucasian; median follow-up of 28 months). It reiterated the established efficacy of HRT (hormone replacement therapy) and SERMs (selective estrogen receptor modulator) [RRR (relative risk reduction) 40%–44%, tibolone, raloxifene], BPs [(RRR 31%–56%), ALN, RIS, ZOL, and IBAN], Dmab (RRR 68%), anabolics (RRR 74%–87%, TPTD, abaloparatide) in reducing the risk of vertebral fractures. Similarly, BPs (ALN, RIS, ZOL), Dmab, HRT (estrogen with or without progestogen), and a calcium/vitamin D combination reduced the risk of hip fractures by 40%. BPs (ALN, RIS, ZOL), Dmab, TPTD, abaloparatide, hormone therapy (HT), tibolone, calcium or vitamin D, romosozumab, and bazedoxifene reduced the risk of nonvertebral fractures by 17%–46%. The second systematic review revealed that women gave equal weight to effectiveness and adverse events, followed by the convenience of taking the drug. Cost and duration of treatment were less important factors for decision-making. The CPG committee developed an approach regarding BP drug holidays based on risk stratification. Low risk included subjects with no prior hip or spine fractures, a BMD T -score at the hip and spine both above −1.0, and 10-year hip fracture risk <3%, or 10-year risk of major osteoporotic fractures <20%; moderate-risk subjects with no prior hip or spine fractures, a BMD T -score at the hip and spine both above −2.5, or 10-year hip fracture risk <3%, or risk of major osteoporotic fractures <20%; high-risk subjects as those with a prior spine or hip fracture, or a BMD T -score at the hip or spine ≤−2.5, or 10-year hip fracture risk ≥3%, or risk of major osteoporotic fracture risk ≥20%; and very high risk includes multiple spine fractures and a BMD T -score at the hip or spine equal to or below −2.5 . The guidelines recommended BPs as first-line therapy to treat postmenopausal women at high risk (high-quality evidence), and for those taking BPs, that fracture risk be reassessed after 3–5 years, and that women who remain at high risk of fractures to continue therapy, while those who are at low-to-moderate risk of fractures to be considered for a “BP holiday (low-quality evidence).” BP holiday was defined as a temporary discontinuation of the drug for up to 5 years, possibly longer depending on BMD and clinical scenario. The guidelines recommend reassessment of fracture risk every 2–4 years and consideration of osteoporosis therapy reinitiation earlier than the 5-year holiday maximum, in the event of an interim fracture, a significant decline in BMD, or increased clinical risk status. The CPGs also underscored that fracture risk stratification should be determined using country-specific assessment tools to guide decision-making. Its derived cutoffs were based on FLEX and HORIZON extension studies that consisted mostly of Caucasian subjects, and the T -score cutoff and thresholds specified and may not be applicable to other populations.
79.4.3
American Association of Clinical Endocrinologists 2019: clinical practice guideline for the diagnosis and treatment of postmenopausal osteoporosis
American Association of Clinical Endocrinologists (AACE) issued their original set of recommendations for the management of postmenopausal osteoporosis in 2010. The original document suggested a drug holiday after 4–5 years of BP treatment in patients at moderate risk of fractures, and after 10 years for high-risk patients, but the terms high and moderate risk were not defined . The update includes an additional risk stratum: very high-risk patients . In the 2019 update, AACE refined risk stratification as follows: Recommendation 23. Patients at very high fracture risk include those with a recent fracture (e.g., within the past 12 months), fractures while on approved osteoporosis therapy, multiple fractures, fractures while on drugs causing skeletal harm (e.g., long-term glucocorticoids), very low T-score (e.g., <−3.0), high risk for falls or history of injurious falls, and very high fracture probability by FRAX® (fracture risk assessment tool) (e.g., major osteoporosis fracture >30%, hip fracture >4.5%) or other validated fracture risk algorithm. Patients, who have been diagnosed with osteoporosis but are not at very high fracture risk, as defined above, are considered high risk (Grade B; BEL 1; downgraded due to limited evidence). With regard to drug holidays in patients on BPs, the 2019 update recommends the following: Recommendation 36 . For BPs, consider a BP holiday after 5 years of oral treatment, or 3 years of intravenous treatment, if fracture risk is no longer high ( T -score >−2.5, no fractures, etc.), but continue treatment if fracture risk remains high (Grade B; BEL 2): Recommendation 37. For oral BPs, consider a BP holiday after 6–10 years of stability in patients with very high fracture risk and continue treatment if fracture risk remains high (Grade B; BEL 2): Recommendation 38. For ZOL, consider a BP holiday when fracture risk is no longer high, but continue treatment when fracture risk is high (Grade A; BEL 1): Recommendation 39. The ending of a BP holiday should be based on individual patient circumstances such as an increase in fracture risk, a decrease in BMD, or an increase in BTMs (Grade A; BEL 1). For information on grading of the evidence, please refer to Table 79.1 .
![](https://clinicalpub.com/wp-content/uploads/2023/09/banner1.png)