At the 2017 San Antonio Breast Cancer Symposium (SABCS), the International Breast Cancer Study Group (IBCSG) and its collaborators presented the 8-year updates of the key modern trials of ovarian function suppression after local treatment for young women with resected breast cancer.1 These updates of the SOFT and TEXT trials were just published in The New England Journal of Medicine.2
For the first time, the answer to the primary question posed by SOFT was positive: Ovarian function suppression given for 5 years does indeed reduce disease-free survival events when added to 5 years of oral adjuvant tamoxifen. Disease-free survival at 8 years went from 78.9%
to 83.2%, a hazard ratio (HR) of 0.76 with P = .009. One-third of women entered on the SOFT trial were randomized to receive the aromatase inhibitor exemestane plus ovarian function suppression—they had an even better disease-free survival (85.9% at 8 years, the median follow-up).
Only 54% of disease-free survival events in the SOFT trial involved distant metastases. The latter are almost always the mechanisms of disability and death in breast cancer. Overall survival is the best endpoint for major and toxic therapies applied to healthy patients, especially young and vigorous ones. If a surrogate endpoint is needed to provide early indications of the ultimate overall survival results, the best one is distant disease–free survival. The SOFT trial has provided us with invaluable data regarding distant recurrence–free interval, which censors deaths without identified distant recurrence rather than counting it as a negative event, as it would be if the endpoint were distant recurrence–free survival. When we counsel patients faced with choosing therapy, the inclusion of death from any cause (where cause of death is not specified) makes distant recurrence–free survival the preferred endpoint. From the point of view of the patient, death is always undesirable.
Primary Analysis Now Positive but Was Negative in 2015
Despite concerns (mostly on the part of this author) that improved disease-free survival is not the best endpoint for an adjuvant trial of toxic systemic therapy in resected breast cancer, that is the primary endpoint that was chosen before the trials began in 2003, and disease-free survival has been precisely defined by the international consensus STEEP conference.3 We should be pleased to have moved from the shaky position of doing subgroup analyses looking for positive effects in a negative trial to the justifiable one of examining the results of a positive trial to decide whether the results are of sufficient magnitude to justify their application to patient groups with varying risks of distant metastasis and death from breast cancer. This is a much better position from the point of view of valid statistical analysis.Error loading Partial View script (file: ~/Views/MacroPartials/TAP Article Portrait and Quote.cshtml)
We knew before SOFT that ovarian function suppression by itself delayed relapse of and death from receptor-positive breast cancer in premenopausal women. SOFT answered the important question of whether ovarian function suppression adds to the benefits of tamoxifen and chemotherapy in this population.
The point of this editorial is not to contest the finding, but to discuss to whom it should be applied, how it fits in with new knowledge about endocrine therapy after 5 years of tamoxifen for these women, whether SOFT and TEXT included groups of patients we now analyze and treat separately (those whose tumors have amplified HER2), and whether we have sufficient information to allow us to start endocrine therapy for these young women with ovarian function suppression and an aromatase inhibitor.
The exemestane issue remains unsettled by the data now available. The overall absolute survival benefit for ovarian function suppression plus tamoxifen plus chemotherapy at 8 years was 4.3 % over tamoxifen alone, but it was only 2.1% if exemestane was substituted for tamoxifen. Exemestane seems inferior for the “hard” endpoint of overall survival, but it was superior at preventing distant metastases at 8 years by 7% in SOFT and 5% in TEXT among HER2-negative chemotherapy-treated patients.
This editorial will review the data and the issues. Hopefully a few more years of follow-up will make a decision on ovarian function suppression easier for breast cancer patients and their physicians.
IBCSG Conclusions Now Much More Nuanced
When the SOFT and TEXT trials were presented in 2014, the conclusions were that ovarian function suppression was good, especially with exemestane, and oncologists should go and do it, at least in higher-risk patients. In 2017, the conclusion of the combined analysis of the SOFT and TEXT presentations was, “Oncologists need to discuss and weigh benefits and toxicity in each individual hormone receptor–positive premenopausal patient.” The cautious tone is justified by the data and was not challenged in San Antonio.
‘Low-Risk’ Women Still Do Very Well on Tamoxifen Alone
Both the SOFT and TEXT trials allowed the entry of women judged on clinical grounds not to require adjuvant chemotherapy. These women did very well in the prior 5-year analysis, and they continue to do very well in the current 8-year analysis for SOFT and 9-year analysis for TEXT. The SOFT and TEXT patients were older and did better than expected when the studies were designed. One suspects, though, if the designers of these two trials had known how much the outcome of these women would improve with the turn of the 21st century, they would not have included them in the trial. It turns out that the excellent outcomes in women considered sufficiently low risk not to require chemotherapy diluted the benefits of the interventions being studied!
The inclusion of women with hormone receptor–positive HER2-positive tumors was appropriate when the SOFT and TEXT trials were designed—it only makes problems for us now!— Steven E. Vogl, MD
Tweet this quote
The 8-year overall survival of these women in the SOFT trial was 98.8% for tamoxifen, 97.9% for tamoxifen plus ovarian function suppression, and 97.7% for exemestane plus ovarian function suppression. Half of the 24 deaths among these 1,419 women were without prior distant metastases, and so they were likely not related to breast cancer. The 8-year rate of being free of distant disease was 97.8% for tamoxifen, 97.8% for tamoxifen plus ovarian function suppression, and 99.3% for exemestane plus ovarian function suppression. Eight-year disease-free survival among these women went from 87.4% for tamoxifen to 90.6% for tamoxifen plus ovarian function suppression to 92.5% for exemestane plus ovarian function suppression. Clearly, only a few of the disease-free survival events prevented or delayed by ovarian function suppression with or without using exemestane instead of tamoxifen were life-threatening.
Most experts have concluded that these “low-risk” women do not need more adjuvant therapy than tamoxifen alone. The argument is very strong because the overwhelming majority of these women did very well, and it makes no sense to subject 99 women to the ravages of an early menopause to benefit 1 woman with delayed distant metastases. One suspects that these very low–risk women would derive little or no benefit from chemotherapy either (as was noted in the TAILORx study).
We do not know exactly how the physicians and the patients chose those women to avoid chemotherapy, but we know they turned out to be older, had smaller primary tumors that were better differentiated, with few or no positive lymph nodes, and had cancers that were almost never HER2-positive. Although we cannot exactly reproduce the criteria used to choose “no chemotherapy,” since they were not spelled out in the protocol, we can probably come pretty close using these clinical characteristics, as well as markers of proliferation such as Ki67 and gene-expression profiles such as Oncotype DX and MammaPrint.
Relevant Analysis Involves Chemotherapy-Treated Patients
We are left deciding whether the benefits to the higher-risk patients (those selected to receive chemotherapy) are worth the toxicities. In the SOFT trial, the benefit of adding ovarian function suppression to tamoxifen in this population is a 2.1% reduction in 8-year distant events (to 17.9% from 20.0%, HR = 0.84) and a 4.3% reduction in deaths (to 10.6% from 14.9%, HR = 0.59). Using exemestane plus ovarian function suppression did reduce distant events compared with tamoxifen (to 15.5%, HR = 0.74), but the reduction in deaths of 2.1% was smaller (HR = 0.79).
Among chemotherapy-treated patients, exemestane plus ovarian function suppression led to a slightly worse overall survival than tamoxifen plus ovarian function suppression in the SOFT trial, but a slightly better one in the TEXT trial. This may be due to chance, but it is more likely a reflection of study eligibility. Women in the TEXT trial, who entered before chemotherapy was started, were more likely to have permanent ovarian failure from their chemotherapy and to not depend only on the triptorelin (Trelstar, Triptodur) injections to shut down their ovarian endocrine function. Since triptorelin does not always work, a drug like exemestane—which depends on nonsecreting ovaries—will work better in a study where more of the population has permanent ovarian failure without triptorelin. The SOFT trial was designed to demonstrate the effect of ovarian function suppression in women with functioning ovaries. Those with permanent ovarian failure were therefore excluded from SOFT, but not from TEXT (where randomization and initiation of triptorelin, if assigned, was before chemotherapy). The benefits in 8-year disease-free survival were more consistent among SOFT and TEXT, from 71% for tamoxifen to 76.7% for tamoxifen plus ovarian function suppression to 80.5% for exemestane plus ovarian function suppression.
The Really Relevant Analysis Excludes HER2-Positive Tumors
The inclusion of women with hormone receptor–positive HER2-positive tumors was appropriate when the SOFT and TEXT trials were designed—it only makes problems for us now! The big problem is that the disease-free survival benefit to adding ovarian function suppression to tamoxifen is largely driven by the 236 HER2-positive cases among the approximately 2,000 patients in the comparison. The hazard ratio for disease-free survival for adding ovarian function suppression is 0.41 for women with HER2-positive cancers, but only 0.83 for those with HER2-negative cancers (the 95% confidence limit for the latter overlaps 1.0). The test for interaction is positive at P = .04. Given this finding, it is surprising that disease-free survival improvement for exemestane plus ovarian function suppression over tamoxifen was not greater in the HER2-positive population than in the HER2-negative one. I have not heard a convincing explanation for this observation beyond the play of chance.
We should reassure [women with HER2-negative cancers] that the vast majority of women in their situations do well whether or not they receive adjuvant ovarian function suppression….— Steven E. Vogl, MD
Tweet this quote
Since we now segregate HER2-positive cancers and treat them differently, we need to see outcomes for distant recurrence-free interval and overall survival among women with higher-risk disease (who were given chemotherapy) and whose cancers were negative for HER2. For women with HER2-negative cancers in both the SOFT and TEXT trials who received chemotherapy, distant recurrence–free interval favored exemestane plus ovarian function suppression over tamoxifen plus ovarian function suppression, with a hazard ratio of about 0.69 and an 8-year difference of 7% in SOFT and 5% in TEXT. Overall survival among all HER2-negative patients (with or without chemotherapy) favored exemestane plus ovarian function suppression over tamoxifen plus ovarian suppression, with a hazard ratio of 0.86, but the 95% confidence interval includes 1.0, and so the difference is not statistically significant.
The 8-year overall survival advantage for exemestane over tamoxifen (each plus ovarian function suppression) in TEXT was 1.9%, with a hazard ratio of 0.74 that was not significant. In SOFT, the former had a 1% advantage in 8-year overall survival, but the hazard ratio overall was 1.02 favoring the tamoxifen arm, and no significant difference was detected in SOFT either.
Only longer follow-up will tell us whether the encouraging differences in distant recurrence–free interval favoring exemestane plus ovarian function suppression over tamoxifen plus ovarian function suppression translate into a long-term significant advantage in overall survival. Until then, we are left with the surrogate endpoint of distant recurrence–free interval. We have not yet seen analyses of distant recurrence–free interval and overall survival for all three arms of SOFT for the HER2-negative, chemotherapy-treated population. One can guess that the statistician thinks the numbers are too small to justify valid conclusions.
What we have seen is a disease-free survival analysis for the approximately 860 HER2-negative patients whether or not they had chemotherapy, with about 860 women per arm. The advantage for tamoxifen plus ovarian function suppression was not significant, with a hazard ratio of about 0.84. Only 21 disease-free survival events separate tamoxifen plus ovarian function suppression and tamoxifen alone at the median follow-up of 8 years, and it is likely that only about half of these were distant events. With so few distant events separating the arms, we need further follow-up and more distant events to draw firm conclusions.
For a disease-free survival comparison of exemestane plus ovarian function suppression vs tamoxifen among HER2-negative women with or without chemotherapy, 63 events separate the group, with a hazard ratio of about 0.60 and a narrow confidence interval. Again, one can guess that about half of these were likely distant events.
How About Ovarian Suppression for HER2-Positive Cancers?
Analysis of the benefits of ovarian function suppression in the HER2-positive population will be important but complex. Only 61% of the women with these cancers in the SOFT and TEXT trials received adjuvant trastuzumab (Herceptin), the standard of care for adjuvant therapy of HER2-positive, node-positive cancers since 2005. The SOFT statisticians and investigators promise this analysis in the near future. The raw numbers in Table S5 of the supplementary appendix published by Francis et al suggest fewer distant recurrences and fewer deaths with tamoxifen plus ovarian function suppression than with exemestane plus ovarian function suppression, but the numbers are small, and HER2-directed treatments were applied inconsistently.2 Given the small number of subjects and the heterogeneous treatments, there will likely be little definitive to conclude from the analysis.
Considering the plethora of anti-HER2 therapies in development, it seems unlikely a large adjuvant trial of ovarian function suppression in this population will be undertaken in the near future, even though ovarian function suppression may prove cost-effective compared with the outrageous cost of recently approved adjuvant HER2-directed therapies. Further complicating statistical planning for a trial of ovarian function suppression in HER2-positive patients is the already excellent prognosis of these women. For instance, node-positive women with HER2-positive cancers enjoyed a 10-year overall survival of 84% before ado-trastuzumab emtansine (also known as T-DM1; Kadcyla) and pertuzumab (Perjeta) were available, whereas node-negative populations have almost no distant metastases with just 12 weekly paclitaxel doses and 1 year of trastuzumab plus standard endocrine therapy.4,5
No Patients in the SOFT Trial Received Current Optimal Therapy
It was 2013 before it became clear that women who are premenopausal at diagnosis benefit substantially from prolongation of oral endocrine therapy beyond 5 years, either by prolonging tamoxifen or switching to an aromatase inhibitor after intercurrent menopause. How to achieve this benefit for women given 5 years of exemestane plus ovarian function suppression is unclear, since prolongation of an aromatase inhibitor beyond 5 years has not yet been shown to have more than minor advantages, which are largely in the prevention of new primary breast cancers.
A switch to tamoxifen has never been studied after 5 years of aromatase inhibitor therapy. It seems likely that administration of an aromatase inhibitor after chemotherapy has led to an early menopause will at least in part balance the early advantages of initial ovarian function suppression together with either tamoxifen or an aromatase inhibitor—and perhaps even surpass them.
Actual Therapy Often Deviated From the Protocol
Further complicating efforts to apply the analysis of the SOFT trial are the protocol deviations—16% of tamoxifen patients received ovarian function suppression before the 5-year point; 20% of those assigned to ovarian function suppression stopped ovarian function suppression early; assigned oral hormonal therapy stopped early in 23% on tamoxifen alone, 19% on tamoxifen plus ovarian function suppression, and 28% on exemestane plus ovarian function suppression. Such deviations are common both in trials and in the “real world.” Both ovarian function suppression plus tamoxifen and ovarian function suppression plus exemestane would likely have appeared more effective if more women had completed the planned 5 years of therapy, and the tamoxifen control group might have done worse if one in six patients had not had ovarian suppression, which we now know is effective. It remains to say how much it is effective in a modern context.
The reduction in distant recurrences with ovarian function suppression plus exemestane over ovarian function suppression plus tamoxifen among chemotherapy-treated subjects is particularly impressive, considering the large numbers of women who stopped exemestane and ovarian function suppression early. This means that quoting the study results may substantially understate the benefits achieved among compliant patients.
Discuss These Issues With High-Risk Menstruating Women
Unfortunately, even though the SOFT trial is now positive, the discussion has not become easier. We still lack detailed, long-term, significant information on the advantages of ovarian function suppression on distant recurrence–free interval and overall survival in higher-risk women, with HER2-positive patients excluded. Ovarian suppression produces more hot flashes, sexual dysfunction, and osteoporosis and has been associated with long-term effects on cardiac health and excess second cancers in the lung and colon.
The small survival advantage from adding ovarian function suppression to tamoxifen might not have occurred if the control group had continued tamoxifen to menopause and then switched to an aromatase inhibitor. It seems likely that the overall survival advantage was driven by events prevented in women with HER2-positive cancers, as was the case for disease-free survival. We have not yet heard or seen overall survival data favoring tamoxifen plus ovarian function suppression for chemotherapy-treated women with HER2-negative tumors. Only about 25% of tamoxifen-treated women continued hormonal therapy beyond 6 years, and less than 13% of exemestane-treated women did so. By 2018 standards, the tamoxifen control group was undertreated with hormonal therapy.
These major considerations aside, should we use a 2.1% reduction in distant events at 8 years and a 4.3% improvement in overall survival in the higher-risk subgroup that received prior chemotherapy in the SOFT trial to recommend ovarian function suppression plus tamoxifen to our patients? In favor of this strategy, the curves presented at SABCS 2017 and published in The New England Journal of Medicine on June 4 seem to separate more at year 9, suggesting that the differences will enlarge. I will tell my patients that the 4.3% survival difference is likely real, and reflects fewer ultimately lethal relapses in the first 6 years. It seems likely that prolonged endocrine therapy (which is now standard) will shrink the differences in distant recurrence and survival that favor ovarian function suppression in the SOFT trial, rather than letting them grow beyond year 8. Randomized trial data on this point will likely never be available.
Since we will have this discussion only with those with HER2-negative cancers, we have to tell them that we do not yet have the appropriately restricted analyses. I am uncomfortable presenting this complicated issue to worried young women facing a potentially lethal disease, but I see no ethical alternative. We should reassure them that the vast majority of women in their situations do well whether or not they receive adjuvant ovarian function suppression and probably will do better now than they did 15 years ago, when SOFT and TEXT began to accrue subjects.
The large advantages to exemestane plus ovarian function suppression over tamoxifen plus ovarian function suppression (5%–7%) in preventing distant recurrence have not yet translated to a benefit in overall survival. Perhaps this will develop with further follow-up. Because in adjuvant studies overall survival is the ultimate endpoint, and because there is no hint of the survival curves separating at year 9, I favor telling our patients about the data but not currently recommending exemestane plus ovarian function suppression as adjuvant hormonal therapy. If the as yet unpresented and unpublished data restricting the analysis to women with HER2-negative cancers who were judged to be at high risk (and received chemotherapy) show a major survival difference in favor of exemestane plus ovarian function suppression over tamoxifen alone, the extent and duration of the benefit will likely dictate my recommendations.
SOFT and TEXT Investigators Deserve Our Gratitude
Those who designed the SOFT and TEXT trials and now have twice analyzed the data have done our patients a great service. Ovarian suppression in the treatment of resected breast cancer is unlikely to be examined again soon, so their continued analyses are all we are likely to have on which to base our advice to these worried young women. Despite the evidence that exemestane plus ovarian function suppression prevents more distant metastases, we have documented no survival advantage at 8 years. Tamoxifen plus ovarian function suppression improves survival at 8 years by 4.3%, but this seems largely driven by fewer deaths among women with HER2-positive cancers, whom now we consider separately.
Some of these issues will likely be clarified in updated papers on these two trials in 2 or 3 years. It would simplify our advice to patients if the overall survival advantage for tamoxifen plus ovarian function suppression goes from 4.3% at 8 years to 8% at 12 years, if this extends to those with HER2-negative cancers, and if the impressive distant recurrence reduction with exemestane plus ovarian function suppression among chemotherapy treated HER2-negative patients translates to a 6% survival advantage over tamoxifen plus ovarian function suppression at 12 years. We, and our patients, await the data. ■
Dr. Vogl is an oncologist who works in the community oncology setting. He is affiliated with Montefiore Medical Center in New York City and White Plains Hospital Center.
Disclaimer: The opinions expressed in this column are those of the author. If you would like to share your opinion on this or another topic, please write to editor@ASCOPost.com.
DISCLOSURE: Dr. Vogl reported no conflicts of interest.
2. Francis PA, Pagani O, Fleming GF, et al: Tailoring adjuvant endocrine therapy for premenopausal breast cancer. N Engl J Med. June 4, 2018 (early release online).
3. Hudis CA, Barlow WE, Costantino JP, et al: Proposal for standardized definitions for efficacy end points in adjuvant breast cancer trials: The STEEP system. J Clin Oncol 25:2127-2132, 2007.
4. Perez EA, Romond EH, Suman VJ, et al: Trastuzumab plus adjuvant chemotherapy for human epidermal growth factor receptor 2–positive breast cancer: Planned joint analysis of overall survival from NSABP B-31 and NCCTG N9831. J Clin Oncol 32:3744-3752, 2014.
5. Tolaney SM, Barry WT, Dang CT, et al: Adjuvant paclitaxel and trastuzumab for node-negative HER2-positive breast cancer. N Engl J Med 372:134-141, 2015.