Duration of Aromatase Inhibitor Treatment in Breast Cancer: The Role of the ‘Carryover Effect’

December 15, 2013
Jack Cuzick, PhD

Oncology, Oncology Vol 27 No 12, Volume 27, Issue 12

The so-called “carryover” effect of tamoxifen results in the reduction in recurrence continuing well after treatment has stopped. The carryover effect is seen mostly in the first 5 years after treatment cessation, and recent evidence suggests that after that time, it is largely over.

Estrogen receptor (ER)-positive breast cancer is unique among common solid tumors in that no patient can be considered cured, and recurrence rates remain almost constant for up to 20 years. The recurrence rate is determined by initial endocrine treatment, nodal status, and tumor size, but grade and most molecular markers have little value for predicting late recurrence. In the Arimidex, Tamoxifen, Alone or in Combination (ATAC) trial, women with node-positive or large (≥ 2 cm) tumors had recurrence rates of almost 4%/year for 10 years, with little difference between years 0 through 5, and 5 through 10, while women with node-negative tumors that were smaller than 1 cm had constant annual rates of about 1% over the 10-year period.[1] Grade and molecular markers (such as the Genomic Health, Inc [GHI] 21-gene recurrence score or our immunohistochemical [IHC4] panel) were highly predictive in the first 5 years, but of little value after that.

This high rate of late recurrence underpins the need to consider long-term endocrine treatment. Most research has focused on tamoxifen, for which a Swedish trial clearly showed that 5 years of treatment was more effective than 2 years.[2] Subsequent work has addressed the question of whether 10 years of treatment is better than 5 years. Initial results from the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-14 trial[3] and a Scottish trial[4] suggested little benefit and an increase in side effects with 10 years, leading to the establishment of 5 years as the optimal duration of tamoxifen treatment. However, more recent trials and updates of these earlier findings have not confirmed this, and additional efficacy is now clearly seen for 10 years of tamoxifen compared with 5 years.

Why were the initial findings so misleading? I think there are two major reasons. The first of these is the so-called “carryover” effect of tamoxifen, which results in the reduction in recurrence continuing well after treatment has stopped. The carryover effect is seen mostly in the first 5 years after treatment cessation, and recent evidence suggests that after that time, it is largely over.[5] A full explanation of carryover is difficult. If, for a proportion of patients, treatment completely eradicated disease and provided a “cure,” then carryover would occur but would last indefinitely. Effects on late recurrence due to new disease could be explained by tamoxifen’s effect on slowing the progression of late-stage precursor lesions, leading to a transient delay in new cancers that would last until earlier-stage lesions had progressed to cancer. Although important for preventive therapy,[6] this putative sequence of events seems likely to make only a minor contribution to metastatic disease, and one is left with the need to postulate some temporary effect of tamoxifen on latent micro-metastases.

Thus, when comparing 5 years vs 10 years of tamoxifen treatment, differences in recurrence in years 5 through 10 reflect the effect of active treatment compared with the carryover effect of the previous treatment, thereby diluting any impact of extended treatment. Only in years 10 through 15, when the carryover effect of 10 years of treatment is compared with essentially no effect from 5 years of treatment, do the differences become more clearly apparent.

A second reason for delayed acceptance of longer-duration tamoxifen treatment is the focus on survival as the final arbiter of efficacy. While undeniably important for cancers with relatively good outcomes, such a focus can lead to substantial dilution of treatment efficacy and delay in the availability of trial results. More than half of all deaths in trials of ER-positive breast cancer occur without evidence of recurrence, and these clearly are unrelated to disease. To compensate for this 50% dilution in observed effect size, 4 times as many patients need to be randomized to achieve the same power. Even if deaths were restricted to those following recurrence, most patients now live, on average, about 5 years after a recurrence, so substantially longer follow-up is needed to see an effect. This delay is clearly seen in the Early Breast Cancer Trialists’ Collaborative Group meta-analysis, in which the effects of 5 years of tamoxifen on recurrence were essentially complete after 10 years of follow-up, although mortality reductions continued for at least another 5 years, largely due to the recurrence reductions seen in the previous 5 years.[5] A greater focus on distant recurrence as the primary outcome measure would thus facilitate an earlier, but still reliable, assessment of treatment impact.[7]

In contrast to the situation with tamoxifen, there is little direct evidence on the appropriate duration of aromatase inhibitor (AI) treatment, and most judgments are based on an analogy with tamoxifen. Use of AIs after 5 years of tamoxifen is clearly beneficial, but this is at least in part due to their greater overall efficacy, and sheds little light on the question of the optimal duration of AI treatment per se. Most current duration trials are investigating the use of 5 years of an AI after 5 years of any endocrine treatment, and the crucial patient subset for evaluating duration will be after 5 years of an AI.

At the other end of the spectrum is the possibility that a shorter duration of AI treatment may be adequate for some low-risk patients. The finding from the Breast International Group (BIG) 1-98 trial that 2 years of letrozole followed by 3 years of tamoxifen appeared to be as effective as 5 years of letrozole raises the possibility that, at least for low-risk ER-positive patients (eg, node-negative, tumor < 1 cm and screen-detected), the appropriate duration of AI treatment could be less than 5 years. A trial of 5 vs 2 years of an AI (without subsequent tamoxifen) in this low-risk group would be of value.

For higher-risk disease, we need better markers to more clearly identify which patients are likely to benefit from extended treatment. Nodal status and tumor size are clearly prognostic, but grade and many molecular markers are not. Exceptions may be the PAM50 assay[8] and the Breast Cancer Index (BCI) score,[9] both of which have demonstrated some predictive power with regard to late recurrence. Although currently not adequately sensitive, identification of circulating tumor DNA at the 5-year point may one day be able to directly identify residual disease, and the development of markers based on material taken at the 5-year point is clearly an important area for research. A bigger challenge will be to identify markers predictive of response to late treatment.

Financial Disclosure:Dr. Cuzick is the principal investigator for the International Breast Cancer Intervention Study (IBIS)-II trial, which is partially funded by AstraZeneca. He also serves on the speakers’ bureau for AstraZeneca.


1. Sestak I, Dowsett M, Zabaglo L, et al. Factors predicting late recurrence for estrogen receptor-positive breast cancer. J Natl Cancer Inst. 2013; 105:1504-11.

2. Randomized trial of two versus five years of adjuvant tamoxifen for postmenopausal early-stage breast cancer. Swedish Breast Cancer Cooperative Group. J Natl Cancer Inst. 1996;88:1543-9.

3. Fisher B, Dignam J, Bryant J, et al. Five versus more than five years of tamoxifen therapy for breast cancer patients with negative lymph nodes and estrogen receptor-positive tumors. J Natl Cancer Inst. 1996;88:1529-42.

4. Stewart HJ, Prescott RJ, Forrest AP. Scottish adjuvant tamoxifen trial: a randomized study updated to 15 years. J Natl Cancer Inst. 2001;93:456-62.

5. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence
and 15-year survival: an overview of the randomised trials. Lancet. 2005;365:1687-717.

6. Cuzick J, Sestak I, Bonanni B, et al; for the SERM Chemoprevention of Breast Cancer Overview Group. Selective oestrogen receptor modulators in prevention of breast cancer: an updated meta-analysis of individual participant data. Lancet. 2013;381:1827-34.

7. Cuzick J. Primary endpoints for randomised trials of cancer therapy. Lancet. 2008;371:2156-8.

8. Dowsett M, Sestak I, Lopez-Knowles E, et al. Comparison of PAM50 risk of recurrence score with Oncotype DX and IHC4 for predicting risk of distant recurrence after endocrine therapy. J Clin Oncol. 2013;31:2783-90.

9. Sgroi DC, Sestak I, Cuzick J, et al. Prediction of late distant recurrence in patients with oestrogen-receptor-positive breast cancer: a prospective comparison of the breast-cancer index (BCI) assay, 21-gene recurrence score, and IHC4 in the TransATAC study population. Lancet Oncol. 2013;14:1067-76.