Prolonged bisphosphonates and increased fracture risk???

​an analysis on the Women's Health Initiative found that long-term oral bisphosphonate therapy in older women put them at increased risk of a fracture, vs shorter-term therapy (see Drieling RL et al. J Am Geriatr Soc 2017; 65: 1924).

Details:

--in a subgroup of 5120 women with estimated 5-year hip fracture risk ≥1.5% and ≥2 years of self-reported bisphosphonate use, investigators assessed risk for incident fractures following various durations of bisphosphonate therapy (2 years, 3–5 years, 6–9 years, and 10–13 years)

--mean age 80 (97% were >70yo); 94% white​; 80% post-high school education; 54% prior fracture after age 54; 8% rheumatoid arthritis; 4% on steroids;  3% on >2 servings ETOH/d; 4% current smokers; 50% very good or excellent health/40% food health; 10% diabetes; 50% estrogen use 6-10 yrs before; 92% on calcium supplements; 14% cancer

--642 (13%) used bisphosphonates for 2 years, 1746 (34%) for 3-5 yrs, 1031 (20%) for 6-9 yrs, and 1701 (33%) for 10-13 yrs.

Results:

--over 3.7 yrs of follow-up, there were 127 hip fractures, 159 wrist or forearm fractures, 235 clinical vertebral fractures, and 1313 "all clinical fractures"

--unadjusted overall fracture risk was lowest in those with only 2 yrs of bisphosphonates, and highest in those on 10-13 yrs

--compared with 2 years of bisphosphonate use, multivariate-adjusted risk:

   --10 to 13 years of use was associated with a modestly but statistically significantly higher risk for any clinical fracture (hazard ratio, 1.29; 95% confidence interval, 1.07–1.57). 

   --3 to 5 and 6 to 9 years' use did not confer excess fracture risk overall or at specific sites.

--the above risks were quite similar in those 2,779 women who had a history of fracture after age 54: HR 1.30 (1.01-1.67)

--Risks after 10 to 13 years of use were even higher for hip fractures (HR, 1.66) and clinical vertebral fractures (HR, 1.65), although these did not reach statistical significance, possibly because of the small numbers in these subgroups. 

Commentary:

--the epidemiology of fractures is pretty scary: age is the main risk factor, 50% of women get a fracture after age 50, low bone mineral density (BMD) increases fracture risk, and by 2020 about 61 million US adults will have low BMD

--the main therapy at this point is bisphosphonates, which increase BMD by inhibiting bone resorption in the bone remodeling process. my review, however, is that the full mechanism is still not clear, and that there seems to be clinical protection even when the BMD does not increase, suggesting that there may be structural improvements in the bone that prevent fractures.  BMD is a gross quantification of the mineral content of the bone and reveals nothing about the microstructure.

--in 2011 the FDA questioned the necessity for long term bisphosphonates, suggesting that patients be routinely evaluated for the appropriateness of continued therapy.  this recommendation reflected both some evidence of harm (atypical femoral neck fractures, osteonecrosis of the jaw) and evidence from the FLEX trial:

    --the FLEX trial randomized 1099 women who had been on alendronate for 5 years to alendronate 5-10mg/d vs placebo for another 5 years, finding that those on placebo had more of a gradual decline in their BMD and increase in their markers for both turnover (though all better than their initial baseline), and there was no difference in nonvertebral or morphometric vertebral fractures by xray (though there were more clinical vertebral fractures  (5.3 vs 2.4%), which was statistically significant (see Black DM. JAMA 2006; 296: 2927). of note, they excluded women who had baseline T scores < -3.5, or whose scores at 5 years were less than in the beginning. post-hoc analysis found that those 720 women without a vertebral fracture at baseline did have fewer nonvertebral fractures if there femoral neck T-score was -2.5 or less. 

    --another trial, the HORIZON study, looked at 1233 women on IV zoledronic acid for 3 years, then stopping vs continuing for 3 years vs placebo, finding fewer morphometric vertebral fractures in those continuing the med (3% vs 6.2%), but no diff in other fractures. the 9-year extension of the trial found only minor changes in bone turnover markers and hip BMD, suggesting stopping 3 years earlier seemed to maintain zoledronic acid benefits.  it was not clear in this study what the baseline BMDs were, though they included postmenopausal women with T-scores less than -2.5. were there many women with T scores less than -3.5???

--in terms of the severe but rare adverse effects:

    --osteonecrosis of the jaw: occurs in 1 in 10K to 1 in 100K patient-yrs, more commonly in those with cancer, altered immune systems, and on very high IV doses

    --atypical femoral fractures: rare complication, mostly associated with longer-term bisphosphonates (median 7 yrs), approx 3.2-50 cases in 100K patient-yrs, but may increase with longer duration of therapy

--the main concern in such an observational study is: were there unreported or unrecognized differences in the women who were on bisphosphonates longer?? did they have increased susceptibility to fractures, more than those on shorter courses, and even the bisphosphonates could not protect them? perhaps they would have had even more fractures if their bisphosphonates were stopped earlier?? (we do not even know the T-scores of the participants). in this regard, the randomized FLEX trial is particularly useful in finding that although there was some deterioration of BMD in those who stopped alendronate, there was no significant increase in fractures overall (though there were more clinical vertebral fractures on stopping therapy), for at least for the 5 year period off meds. but there are no longer term studies, when and if we should restart meds, or if we should continue with a lower dose of bisphoshonates (the FLEX trial did have alendronate at 5mg and 10mg, but primary analysis was by pooling the groups, and there was no comment on the groups separately, though one of their tables did show some difference in BMD favoring the higher dose. no comment on fracture risk). lots of unresolved questions. 

so, this study does add a few things: 

    ​--overall, contrary to the one-liner summary in Journal Watch that "longer is not necessarily better", this study (to me) actually suggests that it is pretty safe to continue the meds for up to 13 years, since the increased risk of fracture even in this study was not apparent until 10-13 years after starting, was not very large, and was not statistically significant for any specific site. And, the selection bias in this study, if any,​ would be that people on bisphosphonates longer would be the ones most susceptible to fractures.

    --and one advantage of this study is that they  started with elderly women, looked at outcomes many years later (into their 90s)

    --that being said, as with all drugs, it is best to use them the shortest period of time necessary to get the desired clinical benefit

    --and, as an observational trial, one cannot conclude that longer term therapy is in fact harmful. but we know that fractures are really common and can be.

my approach will continue to be to repeat the BMD after about 5 years (as per FLEX), and if the BMD is at least the same as the initial one, to stop the bisphosphonates and recheck the BMD in 2-3 years (where a significant decrease might lead me to bisphosphonate reinstatement). but still, even with this study, i would be hesitant to stop bisphosphonates in people with more severe osteoporosis (T-score worse than -3.5), since they are at quite high risk of fracture, the studies of discontinuing meds did not include this group, and the likelihood of the major adverse effects (jaw osteonecrosis and atypical fractures) is quite small. 

 seehttp://gmodestmedblogs.blogspot.com/2016/01/guidelines-on-length-of-bisphosphonate_26.html/​ . for more information on these trials