TAILORx: How to Apply This Landmark Study

Get Permission

Steven E. Vogl, MD

Steven E. Vogl, MD

TAILORx changes the configuration of the ball field and the shape of the ball in deciding which women will be recommended chemotherapy after resection of node-negative, hormone receptor–positive breast cancer. TAILORx was presented by Joseph Sparano, MD, at the 2018 ASCO Plenary Session and simultaneously published in The New England Journal of Medicine (NEJM).1 It is a huge, monumental achievement that casts honor on the investigators who designed, conducted, and analyzed it, as well as the women who participated in it (10,273 of them recruited between 2006 and 2010). Breast cancer patients and their physicians will appreciate the efforts involved for generations and are already grateful for them. The published analysis is honest, critical, and carefully considered. 

Women in TAILORx Did Great! 

LIKE ALL IMPORTANT studies, TAILORx leaves some issues incompletely addressed and others inadequately emphasized. Perhaps in the latter group are the superb outcomes reported even for the highest-risk group. They are summarized in Table 1 (adapted from the NEJM paper).1 

TABLE 1: 9-Year Outcomes in TAILORx

Score (RS)
Freedom From
0–10 84.0% 96.8% 93.7%
11–25 83.3% 94.5% 93.8%
> 25 75.7% 86.8% 89.3%

None of the women with RS < 11 were assigned chemotherapy, all with RS > 25 were, and half with RS 11–25 were randomized to chemotherapy. 

Adapted from Table 2 in Sparano et al.1

The recurrence score (RS) referred to in Table 1 is the weighted sum of the expression of 16 genes—selected (around the year 2000) for their presumed importance to breast cancer prognosis on hormonal therapy—compared to the expression of five reference genes, in an assay designed for paraffin-embedded formalin-fixed cancer tissue. The weighting was designed to allow separation of the prognosis of women with hormone receptor–positive, node-negative breast cancer into three prognostic groups. Without any change to the genes used, the technique, or the weighting, the same recurrence score was later shown to predict benefit from adjuvant chemotherapy in addition to tamoxifen in a subset of participants in the National Surgical Adjuvant Breast and Bowel Project (NSABP) B20 trial who had suitable archived tumor blocks available.

We can now tell our patients with breast cancer without involved axillary nodes that it is quite unlikely (RS > 25) or extremely unlikely (RS < 26) that they will die of breast cancer, assuming they take 5 years of single-agent hormonal therapy (and chemotherapy for those with an RS > 25). Especially for the RS < 25 group, it means they and their physicians should be very conservative in employing other therapies, however superficially attractive. Assuming such a therapy is 50% effective at preventing distant metastases, for the RS < 11 group, one would need to treat 63 patients to help 1; for the RS 11–25 group, 36 patients to help 1; for RS > 25, 15 patients to help.

For the women with an intermediate recurrence score, an analysis of first events showed second primary non-breast cancers (4.3%) to be more common than distant metastases (3.0%), local or regional recurrence (2.1%), and contralateral new breast primary cancer (1.4%), with death without an antecedent cancer event not insignificant at 1.7%. These women with recurrence scores of 11 to 25 should not focus on breast cancer risks alone—deaths without cancer and new non-breast primaries occurred in 6% at 9 years compared to 6.5% for breast cancer events. 

We Still Argue About the Best Endpoint 

AT A WEEKLY breast cancer multidisciplinary conference chaired by the TAILORx principal investigator, Dr. Joseph Sparano, I routinely argue with him that the best endpoint for adjuvant breast cancer trials is overall survival, with survival free of distant metastases as a surrogate whose results are available earlier. I argue that toxic systemic therapies like chemotherapy and ovarian ablation are not justified to prevent treatable and curable conditions like local cancer recurrences in the breast and new contralateral breast cancers. It makes no sense to expect either ovarian suppression or chemotherapy to prevent other primary cancers. Other primary cancers should not be counted as negative endpoints, with the exception of endometrial cancer from tamoxifen and acute myeloid leukemia from chemotherapy. 

Dr. Sparano answers, “Yes, overall survival is important, but surviving with a cancer recurrence or new cancer is not quite the same as surviving cancer-free.” I then answer him, “It is our duty to educate our patients that it is not always worthwhile to prevent undesirable events we can treat and cure later. We would have to treat many patients now to do so, rather than treating just a few later (most of them successfully), when they actually need it.” 

“[TAILORx] is a huge, monumental achievement that casts honor on the investigators who designed, conducted, and analyzed it, as well as the women who participated in it.”
— Steven E. Vogl, MD

Tweet this quote

This argument is not settled in our conference, but I will use distant recurrence–free interval (which censors deaths without distant recurrence, rather than counting them as events, as in distant recurrence–free survival) and overall survival as definitive endpoints in this article because I strongly believe that they are more appropriate endpoints than invasive disease–free survival. They have been precisely defined in the STEEP system.3 I have written extensively on the issue of optimal endpoints for adjuvant systemic trials in breast cancer4 and will not repeat those arguments here. 

Withholding Chemotherapy Is Not Inferior 

THE KEY FINDING of TAILORx is the exclusion of the hypothesis that not giving chemotherapy is inferior to giving it for the RS 11–25 group (6,711 women were in the primary analysis). The rate of freedom from distant recurrence at 9 years was 94.5% with endocrine therapy and 95% with chemoendocrine therapy. Overall survival at 9 years was 93.9% with endocrine therapy and 93.8% with chemoendocrine therapy. 

Heretofore, to recommend chemotherapy or not, we had only an analysis of a subgroup of 134 women in the NSABP B20 trial with a recurrence score of 18 to 30,2 or a revised intermediate recurrence score risk group of 279 from the same study with scores of 11 to 25.5 In each analysis, there were almost twice as many women in the chemoendocrine therapy group as in the tamoxifen-alone group (because there were two arms in B20 with chemotherapy and only one without). Because the numbers of patients were so small, and the numbers of events very small, many of us had to tell our patients that even in the lowest-risk group, we could not exclude a distant recurrence benefit of a few percent. Not any more! 

Two important issues have been largely resolved by this analysis. The first is the argument that only gentle chemotherapy was used in B20 (CMF [cyclophosphamide, methotrexate, fluorouracil (5-FU)] or MF [methotrexate and 5-FU]), where no chemotherapy benefit was apparent for women with recurrence scores of 18 to 30. Perhaps more aggressive, more modern chemotherapy would work much better. In TAILORx, 56% of the chemotherapy given to women with a recurrence score of 11 to 25 was a taxane-plus-cyclophosphamide regimen, 29% was anthracycline-based (presumably AC [doxorubicin and cyclophosphamide]; CAF [cyclophosphamide, doxorubicin, and 5-FU]; or similar regimens using epirubicin). Only 7% got both a taxane and an anthracycline (like TAC [docetaxel, doxorubicin, and cyclophosphamide] or AC-T [doxorubicin and cyclophosphamide followed by docetaxel]), and only 7% got CMF. We can safely conclude that more intensive chemotherapy is not beneficial for these women. 

The second issue largely settled by the TAILORx analysis is the applicability of a curve sent in the past with each report of the recurrence score by Genomic Health, the company providing the Oncotype DX test. This curve, first published by Paik et al in 2006,2 smoothed the benefit of chemotherapy observed in the B20 subset analysis through the intermediate-risk recurrence score, suggesting that the benefit of chemotherapy (MF or CMF) increased as the recurrence score increased in the range of 18 to 30. There was no evidence for this in the B20 analysis, nor is there now any in the final analysis of TAILORx. Indeed, when adjusted for clinical covariates (tumor size and grade), endocrine therapy alone had lower rates of distant recurrence than chemoendocrine therapy at the higher recurrence scores in the 11 to 25 range (Figure S6 in the NEJM paper).1 Hopefully, Genomic Health will update its report to reflect this new information. 

“The key finding of TAILORx is the exclusion of the hypothesis that not giving chemotherapy is inferior to giving it for the RS 11–25 group.”
— Steven E. Vogl, MD

Tweet this quote

Women Under 51 With Scores of 16 to 25 Had Some Chemotherapy Benefit 

HAVING ESTABLISHED noninferiority for endocrine therapy alone, Dr. Sparano went looking for a subgroup in which adding chemotherapy was beneficial. This has no statistical validity in generating firm conclusions, because a large number of subset analyses is possible, and some, by chance alone, will show significant differences that will prove nonreproducible. Nonetheless, the analysis is well intentioned and supported by the relatively large numbers of subjects in each subgroup. Tables 2 and 3 give the differences in overall survival and distant recurrence–free interval for women under 51 with recurrence scores of 16 to 20 and 21 to 25. There are bigger advantages to adding chemotherapy if the endpoint chosen is invasive disease–free survival. However, I think the endpoints in Tables 2 and 3 are more relevant, as noted above. 

TABLE 2: Overall Survival at 9 Years for Women < 51 Years Old 

Recurrence Score Endocrine
Therapy Alone
Chemoendocrine Therapy
< 11 98.6%
11–15 96.8% 97.5%
16–20 95.8% 96.1% (confidence limits overlap)
21–25 92.7% 93.9% (confidence limits overlap)
> 25 92.4%

Adapted from Table 3 in Sparano et al.1

TABLE 3: Freedom From Distant Recurrence at 9 Years for Women < 51 Years Old 

Endocrine Therapy Alone Chemoendocrine -Therapy
< 11 98.5%
(n = 801)
97.2% 98.0%
(n = 923)
93.6% 95.2% (confidence limits overlap)
(n = 492)
86.9% 93.4% (confidence limits do not overlap)
> 25 88.7%

Adapted from Table 3 in Sparano et al.1 

Dr. Sparano thinks we should tell our younger patients with a recurrence score of 16 to 25 about these exploratory, hypothesis-generating results. I cannot but agree. For overall survival, the data are not persuasive at all, but for distant recurrence–free interval in the RS 21–25 subgroup, the difference of 6.5% is suggestive. I will tell my patients that the analysis is not statistically valid, but it is based on 492 randomly assigned women, is the only analysis available, and may be correct. We will not have another randomized study on this issue in my lifetime. It might be useful to look at the incidence of distant recurrences within this group by tumor size, grade, proliferation, and type of cancer (lobular vs ductal)—a more granular analysis to generate more hypotheses. 

The seminal study based on archived tissue from the B20 trial2 is of little help. It compares 10-year distant recurrence in women with recurrence scores of 18 to 30 who were assigned tamoxifen (7 events among 45 patients) or tamoxifen plus chemotherapy (9 events among 89 patients). The populations and numbers of events remain far too small for analysis by individual characteristics. 


TAILORx (Trial Assigning Individualized Options for Treatment) enrolled 10,273 women with hormone receptor–positive, HER2-negative, axillary node–negative breast cancer. Of them, 6,711 had midrange recurrence scores of 11 to 25 on the 21-gene Oncotype DX assay and were the subject of this analysis. Researchers evaluated whether these women could be spared chemotherapy or treated as effectively with endocrine therapy alone. All were treated with a taxane- or anthracycline-containing regimen, 90% of postmenopausal women received aromatase inhibitors, and 15% of the premenopausal group underwent ovarian suppression. These 6,711 women were then randomly assigned to receive endocrine therapy alone or endocrine therapy plus chemotherapy.

The study met its primary endpoint, showing noninferiority in the intention-to-treat population. Endocrine therapy was also noninferior for distant recurrence–free interval, recurrence-free interval, and overall survival. For more information, see Sparano JA, Gray RJ, Makower DF, et al: Adjuvant chemotherapy guided by a 21-gene expression assay in breast cancer. N Engl J Med 379:111-121, 2018.

There is limited useful information on this issue from two observational studies—one study from a health plan in a small country and a single-center review from a major cancer center. A third study from the Surveillance, Epidemiology, and End Results registry was a premature analysis at 5 years of a suspect endpoint (breast cancer–specific survival) and will not be considered further here. The University of Texas MD Anderson Cancer Center report of 549 women with recurrence scores of 11 to 25 does not break down advantage for chemotherapy by age,6 nor does the Israeli Clalit Health Services report of outcomes, which had 113 women under 50 with recurrence scores of 18 to 30 and only 3 distant recurrences among them, with or without chemotherapy.7 

One explanation of why younger women with high-intermediate recurrence scores derive benefit from chemotherapy whereas older women do not relates to the known endocrine effect of chemotherapy containing cyclophosphamide—it produces early menopause. Against this argument is the finding that young age was associated with improved chemotherapy benefit for intermediate scores, but menopausal status was not. From the very first reported randomized trials of adjuvant chemotherapy for resected breast cancer published in 1975, younger women benefited from chemotherapy more. 

What About Those With a Score of 26 to 30? 

DR. SPARANO believes women with a recurrence score of 26 to 30 should be encouraged to take adjuvant chemotherapy. The expansion of the high-risk group down to a recurrence score of 26 for TAILORx from the breakpoint of 31 in the original Oncotype DX analysis was done for 3 reasons.5 The first was to reduce the distant recurrence rate for the new intermediate-risk group on tamoxifen alone to 5% at 10 years (based on the B20 experience). This was considered an acceptable rate to allow randomization to tamoxifen without chemotherapy with equipoise. The second was that very few HER2-positive patients had a recurrence score under 26—they were systematically excluded from TAILORx. The TAILORx population was expected to be systematically at lower risk by virtue of the exclusion of HER2-positive patients. Third, the intergroup committee that designed the trial wanted to be sure not to deny chemotherapy to women who might benefit from it, noting that higher recurrence scores were associated with more benefit from chemotherapy. 

The lower end of the intermediate-risk group was moved down to 11 because this was where a smoothed curve started to show a benefit for adding chemotherapy to tamoxifen, though there was overlap in the confidence limits for the two therapies at levels below a recurrence score of 35. One should note that this is the smoothed curve that has essentially been discredited by the just-published data from TAILORx. 

If one accepts the observation that women less than 51 years old with a recurrence score of 21 to 25 benefit from chemotherapy with reduced distant recurrence, and one is willing to act on it, then it is easy to extend this decision to women less than 51 years old with a recurrence score of 26 to 30—if a woman with a recurrence score of 25 needs chemotherapy in addition to tamoxifen, then a woman with a score of 30 needs it more. This extension does not help at all with the majority (more than two-thirds) of women with scores of 26 to 30 who are aged 51 or older. 

The Clalit observational study from Israel had 171 women with a recurrence score of 26 to 30 equally divided between those who did or did not get chemotherapy in addition to hormonal therapy.7 The chemotherapy group had six distant recurrences; the one without chemotherapy had four. Although the assignment to chemotherapy was not randomized, this does not encourage us to recommend using chemotherapy for this group. The MD Anderson observational series did not separately report outcomes with or without chemotherapy for recurrence scores of 26 to 30.

“We can now tell a large majority of women with resected, hormone receptor–positive breast cancer and unaffected lymph nodes that they need not suffer through chemotherapy and its immediate and long-term toxicities.”
— Steven E. Vogl, MD

Tweet this quote

The Achievement Stands 

DESPITE THESE quibbles at the margins of what we have learned, TAILORx is a monumental achievement and a major contribution to quality of care and minimization of toxicity. We can now tell a large majority of women with resected, hormone receptor– positive breast cancer and unaffected lymph nodes that they need not suffer through chemotherapy and its immediate and long-term toxicities. We now have the numbers to exclude all but the smallest of benefits from chemotherapy for them. We can likely tell the same women that they do not need ovarian function suppression either and that their hormonal therapy can probably end after 5 years. ■

At Microphone 1 is an occasional column written by Steven E. Vogl, MD, of the Bronx, New York. When he is not in his clinic, Dr. Vogl can generally be found at major oncology meetings and often at the microphone, where he stands ready with critical questions for presenters of new data. 

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 

Acknowledgment: Dr. Vogl would like to thank Dr. Sparano for reviewing an earlier version of this opinion piece and correcting several errors. Any remaining errors are those of Dr. Vogl alone. 

DISCLOSURE: Dr. Vogl reported no conflicts of interest. 


1. Sparano JA, Gray RJ, Makower DF, et al: Adjuvant chemotherapy guided by a 21-gene expression assay in breast cancer. N Engl J Med 379:111-121, 2018

2. Paik S, Tang G, Shak S, et al: Gene expression and benefit of chemotherapy in women with node-negative, estrogen receptor-positive breast cancer. J Clin Oncol 24:3726-3734, 2006

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. Vogl SE: Adjuvant ovarian suppression for resected breast cancer: 2017 critical assessment. Breast Cancer Res Treat 166:1-13, 2017

5. Sparano JA, Paik S: Development of the 21-gene assay and its application in clinical practice and clinical trials. J Clin Oncol 26:721-728, 2008

6. Barcenas CH, Raghavendra A, Sinha AK, et al: Outcomes in patients with early-stage breast cancer who underwent a 21- gene expression assay. Cancer 123:2422-2431, 2017

7. Stemmer SM, Steiner M, Rizel S, et al: Clinical outcomes in patients with node-negative breast cancer treated based on the recurrence score results: Evidence from a large prospectively designed registry. NPJ Breast Cancer 3:33, 2017.