Anastrozole: what the studies measured

Anastrozole has the cleanest prevention dataset in breast cancer and one of the thinnest bodies of independent replication. Both come from the same place. Almost every mineralisation number cited for this compound was produced inside a single trial programme, so a reader who checks the source finds the same 3,864 women again and again, wearing different paper titles.

What was measured on the hormone side

The pharmacodynamics are not in dispute and they are not the interesting part. A 2013 review describes the third-generation inhibitors, which include anastrozole, as causing more than 98 % inhibition of oestrogen synthesis in vivo, and reports no difference in anti-tumour efficacy between anastrozole, letrozole and exemestane.

The clearest published hormone-axis data point comes from a different population. Forty-two boys with gynaecomastia of under one year duration, mean age 13 (standard deviation 1.8 years), received anastrozole 1 mg daily for six months in an open-label study with no placebo arm. Testosterone to estradiol ratios increased significantly, with a concomitant rise in luteinising hormone and follicle-stimulating hormone concentrations, the expected signature of aromatase blockade. Time to peak concentration was one hour, apparent clearance 1.54 litres per hour, terminal half-life 46.8 hours.

That study also reports breast area falling about 63 % and breast volume about 57 % against baseline at P = 0.004, which the paper calls an exploratory aim. Name the population: adolescent males with gynaecomastia, uncontrolled.

The prevention trial, at full follow-up

IBIS-II randomised 3,864 postmenopausal women at increased risk of breast cancer to five years of anastrozole 1 mg per day or matching placebo, 1,920 and 1,944 women, median follow-up 131 months, close to eleven years.

All breast cancer occurred in 85 women on anastrozole against 165 on placebo, hazard ratio 0.51 (95 % CI 0.39 to 0.66), p below 0.0001. The effect was largest inside the treatment period, 35 against 89 cases, ratio 0.39, and persisted afterwards at 50 against 76, ratio 0.64, p = 0.014. Invasive oestrogen-receptor-positive disease fell 54 %, ratio 0.46 (0.33 to 0.65), and ductal carcinoma in situ to 0.41 (0.22 to 0.79).

The outcome usually left out sits in the same paragraph. Deaths from any cause were 69 on anastrozole and 70 on placebo, hazard ratio 0.96 (0.69 to 1.34), p = 0.82, and breast cancer deaths were two against three. The authors state that further follow-up is needed to assess the effect on breast cancer mortality. An 11-year cut in incidence alongside no mortality difference is not a contradiction; it is what the interval supports.

The bone numbers, and why they come from one group

Three years of bone mineral density data in the IBIS-II bone substudy come from 1,410 of the 3,864 participants, stratified by baseline T score; in the osteopenic stratum 150 of 260 women, 58 %, had paired DXA scans. At the lumbar spine, three-year mean change rose 1.1 % (95 % CI 0.2 to 2.1) on anastrozole plus risedronate, against a 2.6 % fall (-4.0 to -1.3) on anastrozole plus placebo, p below 0.0001; at the total hip, -0.7 % (-1.6 to 0.2) against -3.5 % (-4.6 to -2.3), p = 0.0001.

The untreated comparison is the one usually quoted. Among women not randomly assigned to risedronate, 310 on anastrozole and 342 on placebo, spine density fell 4.0 % (95 % CI -4.5 to -3.4) against 1.2 % (-1.7 to -0.7), and total hip fell 4.0 % (-4.4 to -3.6) against 1.8 % (-2.1 to -1.4), both at p below 0.0001. In the osteoporotic stratum everyone received risedronate, and there anastrozole did worse than placebo: spine rose 1.2 % (-0.1 to 2.6) on 46 women against 3.9 % (2.6 to 5.2) on 60, p = 0.006; at the hip, p = 0.12. The most common adverse event was arthralgia, 114 anastrozole against 94 placebo in the lowest stratum.

The seven-year analysis comes from the same substudy: 528 women without risedronate had paired five- and seven-year data. After withdrawal, lumbar spine density rose 1.25 % (95 % CI 0.73 to 1.77, P = 0.0004) in women previously on anastrozole against 0.14 % (-0.29 to 0.56) on placebo. At the hip, density was unchanged between years five and seven in the anastrozole group while it continued to fall on placebo, -1.35 % (-1.70 to -0.98). The authors conclude the negative effects on bone density in the preventive setting are partially reversible.

So the three-year loss, the risedronate countermeasure and the post-treatment recovery are all one trial, one substudy, one enrolment. That is not a criticism of a large, well-run trial. It is a warning about how the literature reads: a sentence saying aromatase inhibitors cost four per cent of spinal bone density is quoting 310 women out of 3,864 at elevated breast cancer risk.

The independent check is 68 patients, and it was negative

Two other sources sit next to that. The TEAM Japan substudy randomised 68 patients with hormone-sensitive early breast cancer to tamoxifen, exemestane or anastrozole and measured lumbar spine density plus two turnover markers over two years. Spine density was higher on tamoxifen, but the abstract states the intergroup differences were not significant, p = 0.2521 and p = 0.0753. Tamoxifen did lower both turnover markers at one and two years, p below 0.05, and the authors conclude the bone effect of exemestane and anastrozole may be comparable in Japanese postmenopausal women, which is not the same as comparable with tamoxifen.

The second is a head-to-head against tamoxifen in ductal carcinoma in situ: 2,980 women from 236 centres in 14 countries, 1,449 analysed on anastrozole and 1,489 on tamoxifen, median follow-up 7.2 years, 144 recurrences. All recurrence was 67 against 77, hazard ratio 0.89 (95 % CI 0.64 to 1.23); non-inferiority was established, superiority was not, p = 0.49. Deaths were 33 against 36. Adverse events were reported by 91 % and 93 % of the two arms, with more fractures, musculoskeletal events, hypercholesterolaemia and strokes on anastrozole, and more muscle spasm, gynaecological cancers, vasomotor symptoms and deep vein thromboses on tamoxifen.

Where the fracture numbers come from, and who they cover

The largest independent fracture dataset in this space is not an anastrozole trial at all. It is a patient-level meta-analysis of 7,030 premenopausal women with oestrogen-receptor-positive early breast cancer on ovarian suppression, pooling four randomised trials, median follow-up 8.0 years. Recurrence was lower with an aromatase inhibitor, rate ratio 0.79 (95 % CI 0.69 to 0.90), p = 0.0005, and the benefit sat entirely in years zero to four: rate ratio 0.68 there, an absolute five-year recurrence reduction of 3.2 % (6.9 % against 10.1 %), nothing in years five to nine or beyond year ten.

Fractures ran the other way: 227 of 3,528 women, 6.4 %, against 180 of 3,502, 5.1 %, rate ratio 1.27 (95 % CI 1.04 to 1.54), p = 0.017. Breast cancer mortality showed no difference (1.01, 0.82 to 1.24) and all-cause mortality showed no difference (1.04, 0.86 to 1.27). Endometrial cancer was rare in both arms, seven against fifteen.

That population is plain: women with ovarian suppression for breast cancer. The fracture signal is a real randomised number, and it does not transfer to men or to anyone without a breast cancer indication, because none of those groups was studied.

Where this sits on the shelf

There is no aromatase inhibitor in this catalogue and this page carries no product link, for the reason in the note at the top. The honest neighbour on the same endocrine axis is clomiphene, a different mechanism with a different evidence base; the kisspeptin page covers the upstream axis.

One companion page exists but is not yet reachable at this address: the 05.10 wave carries anastrozole against exemestane, which answers a different question, tumour response rather than what the prevention trial measured. The link resolves on that deploy day and stays declared until then.

What the data do not show

They do not show a bone or fracture figure for anastrozole outside IBIS-II, because the three-year, the risedronate and the seven-year results all come from one substudy. They do not show a mortality benefit, because the eleven-year analysis prints 69 deaths against 70. They do not show that the bone loss is fully reversible, because the authors say partially reversible and the hip did not recover. They do not show an effect in men, because the only pharmacodynamic study here is 42 adolescent boys with no comparator, nor that a non-cancer population loses bone at the same rate, because every density measurement here was made in women selected for breast cancer risk.

What we supply

Nothing on this page is a product of ours and this page recommends nothing. It exists so a reader chasing a bone-density claim about aromatase inhibitors can reach the primary trial.

Research use only. This page summarises published studies for research reference. It is not medical advice, not a protocol, and not a suggestion for human use. Nothing we supply is for human or veterinary use.