There are two articles in last week’s New England Journal of Medicine worth looking at (login required for access): one is a randomized trial of alendronate in glucocorticoid-induced osteoporosis, and the other is a review article of Paget’s disease of bone. I want to review the first of these here (N Engl J Med 2006;355(7):675-684).

For the lay reader, let’s define a few terms.

glucocorticoid (GC): this is a class of steroid medication. The one most commonly used in the USA is prednisone. They are given for a host of diseases, especially those which cause inflammation. Inflammation should not be confused with infection (though the latter can also cause inflammation); the inflammation we’re talking about here is usually a result of some other kind of process, such as autoimmune disease. Rheumatologist use lots of GCs.

alendronate: this is a medication from a class called bisphosphonates. These are used in people who have or are at risk of osteoporosis (bone loss resulting in ‘thin’ bones). Since GCs are a common cause of bone loss and the fractures that can result from it, bisphosphonates are used frequently by rheumatologists to prevent this complication.

alfacalcidol: this is an ‘activated’ form of Vitamin D. Vit D is important for calcium absorption and regulating the deposition of calcium into bone.

The point of this study was to see if alendronate helps to prevent bone GC-induced bone loss in rheumatology patients when compared to alfacalcidol. This is an important issue. There are many patients who take GCs for many different diseases and it’s very well established that they induce bone loss. Low bone mineral density (‘thin’ bones) is a major risk factor for fracture. This may not seem like a big deal but it is. As a big example, hip fractures are a major cause of death and disability in older persons. Why? Not completely understood, but the point is that older people (and even younger folks) getting fractures is a big deal.

I saw this article with a bit of surprise. I thought that the value of bisphosphonates in GC-induced bone loss was pretty well established. I did a very quick literature review, and there are indeed articles out there about alendronate for GC-induced OP. But it appears that these were mostly open label studies, or exensions of previous trials. So maybe there was a need for a good quality trial of concerning this question.

The other thing to address here, is why alendronate was compared to alfacalcidol instead of placebo. the reason for this is that it was felt to be unethical to give patients getting GCs a placebo. Alfacalcidol should at least slow down bone loss if not reverse it, and it can be argued that this is a reasonable treatment for GC-induced OP.

OK, let’s get to the article. [Warning: technical stuff below!]

This was an 18 month study. Subjects were adult patients starting GCs for the first time for a rheumatic disease. One third had polymyalgia rheumatica, one third had rheumatoid arthritis, and the remainder had other rheuamtic diseases. At the beginning of the study, they were anticipated to require CGs for at least 6 months. Subjects were randomized to either alendronate 10 mg daily with alfacalcidol placebo, or alfacalcidol 1 mcg daily with alendronate placebo. The all got at least 1500 mg elemental calcium through diet and supplementation.

Subjects were randomized 1:1, with stratification by age (greater than or less than 50), sex, presence or absence of a vertebral fracture, and initial GC dose (greater than or less than 15mg prednisone daily). This is an attempt to make sure these types of subjects are equally represented in both treatment arms, and is a valid way to assign treatments (though it shouldn’t technically be necessary in a randomized trial of sufficient size).

Subjects had bone mineral density (BMD) measured every 6 months. The primary outcome was % change in L-spine BMD at 18 months. They also kept an eye on incidence of new fractures, both symptomatically and radiographically (L-spine X-Rays every six month).

The study was powered to detect a 5% difference between groups with 40% drop out, giving 100 subjects per group.

Table 1

Figure 1They randomized 200 subjects, 101 to alfacalcidol, 99 to alendronate. About 2/3 of subjects were women and the stratification appears to have been sucessful, i.e. the groups are similar. There were 21 withdrawals/deaths in the alendronate goup, 17 in the alfacalcidol group. There actually was one potentially important difference between groups: prior vertebral compression fractures. There were 14 in the alfacalcidol group, 6 in the alendronate group. This is potentially important as it implies that those getting vit D could have an elevated risk of vertebral fractures at baseline.

The other important point is that this appears to be a completers analysis rather than an intention-to-treat analysis (more about this below).

Results: At 18 months, there was a 2.1% rise in BMD in the alendronate group (95% CI 1.1-3.1), and a 1.9% drop in the alfacalcidol group (0.7-3.1), for a difference of 4.0% (2.4-5.0). There were 8 new vertebral fractures in the alendronate group, 13 in the alfacalcidol group. Hazard ratio was 0.4 (0.1-1.4) with alendronate after adjustment for prevalent vertebral deformities.
There were few major adverse events.

And those are the major results.

Two potentially important points. The first is those prevalent vertebral fractures. I’m glad they provided an analysis adjusted for initial, prevalent fractures; they really had to given the obvious difference between the two groups at baseline. Note that that even though the alendronate goup had close to half the number of new fractures as the other, there was no statistical difference. This is not a surprise as the study was not powered to detect this difference. To show one would have required many more subjects.

The other is the lack of an ITT analysis. It would have been nice to see one, or at least something like a best case / worse case anaylsis with those missing patients treated appropriately (i.e. those missing in the alendronate goups given a low BMD change or even negative change, and those in the alfacalidol group given a positive change). This would have made my confidence in the result greater. I’m unsure why they chose not to do this.

The other thing it would be nice to know about is the difference between age groups. It’s likely a high proportion of fractures occurred in older subjects, but unfortunely we can’t tell from the data presented. We can only assume that both older and younger patients benefitted.

On the other hand, it was nice to see that this was an independantly funded study, i.e. not funded by the drug manufacurer. In my opinion that lessens the chance of bias. I’m inclined to think this is a good study and that it adds valuable information.

So, it looks to me like alendronate is generally a good idea, and prevents bone loss in patients who are started on GCs. This implies that its use will prevent compression and other fractures. This supports our current practice of use and should encourage its continued use.