Limits on Tamoxifen Duration Questioned

SAN ANTONIO--The NCI's clinical alert advising physicians to limitthe use of tamoxifen (Nolva-dex) in early breast cancer to nomore than 5 years may be a "premature judgment" thatwas based on a randomized trial of insufficient size, Prof. RichardPeto, of the University of Oxford's ICRF Clinical Trial ServiceUnit, said at the San Antonio Breast Cancer Symposium.

How long should tamoxifen treatment continue? "Current trialsare too small to provide a reliable answer," Prof. Peto said.Worldwide, he continued, the numbers of women who have been randomizedbetween different durations of tamoxifen are not yet adequate.

The trials of 1 year vs longer treatment involved only about 2,000women; trials comparing 2 years with 5 years of tamoxifen arein progress but have not yet produced mature data; and trialsof 5 years vs longer treatment involve fewer than 2,000 women,and these trials have now been stopped, Prof. Peto said.

"The year 2000 overview will bring together the data fromall of these trials, and then we will have really good evidenceon 5 years vs 2 years of treatment," he said. But even thenthe question of whether continuing treatment for more than 5 yearsproduces additional benefits may be difficult to answer reliably.

"The main trial [NSABP B-14] of 5 years vs 10 years, withmore than 1,000 randomized women, has just been shut down dueto what may, in retrospect, turn out to be a misleading alarmby the data monitoring committee," Prof. Peto said. "So,at the moment, we have no directly randomized evidence on 5 yearsvs longer in the pipeline, unless new studies can be started andmade to work."

One such trial being planned is the Adjuvant Tamoxifen LongerAgainst Shorter (ATLAS) study, funded by the US Army, which aimsto randomize 20,000 women. "That's the appropriate scale,"Prof. Peto said. "We'll find out whether it is possible ornot to perform such a study over the next few years."

In his talk, Prof. Peto reported the preliminary results of therecent Early Breast Cancer Trialists' Collaborative Group (EBCTCG)metaanalysis of tamoxifen trials, which is currently being preparedfor publication. [An updated metaanalysis of the effects of radiotherapyand surgery in early breast cancer appeared last November (N EnglJ Med 333:1444-1455, 1995)].

The metaanlysis involved data on 30,000 randomized women fromapproximately 40 trials of tamoxifen use in early breast cancer,with follow-up now out to 15 years or more.

"The results in the first 5 years involved about 4,000 deaths,with another 2,000 in the next 5-year period, and another 1,000in the next 5 years. In each of these periods, the death ratewas slightly lower among the women who had previously taken tamoxifen,"Prof. Peto said.

Overall, the decrease in mortality attributed to any tamoxifenuse is approximately 5% at 15 years (47% vs 42% survival). "Themain finding in the trials is that a few years of tamoxifen substantiallyprolongs disease-free survival in all categories except perhapsestrogen-receptor-poor women," he said.

Prof. Peto showed that the benefits of tamoxifen accrue both toolder and younger women when the treatment duration is appropriatelylong. With more than 2 years of tamoxifen use, absolute improvementin mortality was about 6% in both women over and under age 50."The idea that tamoxifen doesn't work in women under 50 isa statistical mistake that should be laid to rest," he said.

Overall benefits are seen regardless of nodal status, Prof. Petopointed out. The survival benefit at 10 years with any tamoxifenuse is 3% among node-negative women and 6% in node-positive women."Both figures are highly statistically significant,"he said.

Benefits Increase With Longer Use

On average, Prof. Peto said, the trials in the metaanalysis involvedonly 2 years of treatment "in a disease that obviously hasa natural history that can run over a 10- to 20-year period."He noted that the mortality benefit of tamoxifen in early breastcancer patients is "somewhat greater" in the longer-durationtrials. With 1 year of treatment, the metaanalysis shows abouta 4% benefit, 5% with 2 years of treatment, and about 6% withmore than 2 years.

In terms of non-breast-cancer deaths, the metaanalysis showedno difference between those who received tamoxifen and those whodid not, suggesting that the potential side effects of tamoxifendo not materially affect mortality. "I find it disappointingthat there was no apparent improvement in cardiovascular mortalityamong those taking tamoxifen, because the drug is known to lowercholesterol," he said.

Differences in the incidence of endometrial cancer were smallbut real. Among some 30,000 randomized women, there were 78 endometrialcancer cases in the tamoxifen group vs 21 in the controls.

"Clearly a real difference," Prof. Peto said, "butonly three excess cases per 1,000 women." In terms of deathsfrom endometrial cancer, the excess for tamoxifen users was onlyone per 1,000 (19 endometrial cancer deaths for tamoxifen vs 2in the control group).

He also noted that the metaanalysis has completely refuted anyconcerns about increases in colorectal cancer with tamoxifen use.

Although the incidence of endometrial cancer rises with longertamoxifen use, "so do the benefits," Prof. Peto pointedout. For those receiving an average of 4 years of tamoxifen, thenumber of endometrial cancers rose to 9 per 1,000, but the reductionin contralateral breast cancer was 50%, compared with 20% and30% for 1 and 2 years of tamoxifen use, respectively. "Butit is still uncertain what will be seen with really long-termtamoxifen treatment," he said.