What Does a Low Bone Density Measurement Mean?


Synopsis: Bone loss measured after one year of treatment is not a certain indication that the patient is not responding to therapy.

Source: Cummings SR, et al. JAMA 2000;283:1318-1321.

Cummings and colleagues from the fracture Intervention Trial Research Group analyzed hip and spinal bone mineral density data from the randomized clinical trials assessing the effects of alendronate and raloxifene in postmenopausal women with osteoporosis. They found that women who lost bone mineral density after one year of treatment were likely to gain bone during the second year. For example, 83% of the women taking alendronate whose hip bone density decreased by more than 4% in the first year had increases in the second year of treatment. Those who gained an extreme amount of bone in the first year tended to lose in the second year. Similar results were observed with raloxifene treatment. The more extreme the measurement after one year of treatment, the more likely the next year’s measurement indicated a reversal. The same results were observed in the women receiving placebo treatment. Cummings et al argue that these results illustrate the principle of regression to the mean. Regression to the mean is a phenomenon due to extreme results of measurement being in part due to random error, and therefore, repeat results are usually closer to the mean for a population. Cumming et al recommend that bone treatments should not be discontinued when measurements in the first year indicate loss of bone density.


Not all women will maintain or gain bone density on postmenopausal hormone therapy; in one study, 12% of treated women lost bone despite apparently good compliance.1 In the PEPI 3-year clinical trial, where compliance rates were probably maximal, 4% of treated women lost bone in the spine and 6% in the hip.2 In women who started hormone therapy at menopause, I recommend a bone density measurement in the mid 60s to detect nonresponse. In women who have started treatment later in life, or who have started treatment because a bone density measurement has already indicated the presence of osteoporosis, the above report indicates a repeat measurement one year later is not the best method of management.

First, I believe it is important to check for compliance and with estrogen therapy adequate dosing. Compliance with alendronate is a problem because of the requirement for an empty stomach and upright posture for at least 30 minutes. I check compliance with estrogen by measuring blood estradiol levels. Monitoring the estradiol blood level in postmenopausal women receiving hormone therapy is not as straightforward as it would seem. There are two primary difficulties. First, the clinical assays available differ considerably in their technique and quality (laboratory and antibody variations). Second, the various commercial products represent a diverse collection of estrogenic compounds, ranging from estradiol to unique equine estrogens. What each clinician must do is learn what blood level of estradiol as performed by the local laboratory is associated with the standard doses of hormone therapy (0.625 conjugated estrogens, 1 mg estradiol, 50 mg transdermal estradiol). In our laboratory, this range is 40-100 pg/mL estradiol. Remember that because FSH is regulated by a factor other than estrogen (i.e., inhibin), FSH levels cannot be used to monitor estrogen dosage.

If compliance with medication is not a problem, make sure there isn’t another cause of bone loss, especially due to medications or eating disorders.

• Drugs: Heparin, anticonvulsants, high intake of alcohol.

• Chronic Disease: Renal and hepatic.

• Endocrine Diseases: Excess glucocorticoids, hyperthyroidism, hyperparathyroidism.

• Nutritional: Calcium, phosphorous, vitamin D deficiencies.

If a reason for the apparent lack of response is not detected, I recommend measuring one of the urinary biochemical markers of bone turnover. If the marker level is not in the premenopausal range, then this patient is truly a nonresponder. If the marker level is in the premenopausal range, then the bone density measurement probably reflects the problem reviewed by Cummings et al, regression to the mean. Treatment should not be changed and a repeat bone density measurement a year or two later will probably indicate response. Therefore, I would change treatment only if the marker level is not in the premenopausal range.

The reason why some women fail to respond is unknown. It is further unknown whether these patients will respond to added treatment, such as calcitonin or a bisphosphonate, but it is worth special evaluation, treatment, and surveillance. Until we learn more about this group of women, they should be treated with combination therapy (e.g., estrogen and alendronate), and we should closely monitor their response. But keep in mind that fractures reflect a more complex story than just bone mineral density. It is likely that apparent non-responders as determined by bone mineral density still have some treatment-induced protection against fractures.


1. Hillard TC, et al. Osteoporos Int 1994;4:341-348.

2. PEPI Trial. JAMA 1996;276:1389-1396.

The following statements are true regarding the treatment of low bone density except:

a. Bone density measurements after one year of treatment may not be an accurate indicator of response.

b. A lack of bone mineral density response to treatment indicates an extreme risk of fracture.

c. Extreme results in single clinical measurements confirmation; follow-up results may be relatively normal, illustrating regression to the mean.

d. Blood estrogen levels in women on estrogen therapy can be interpreted only if the clinician uses the same laboratory with known normal ranges.