Reply to K. Ando et al

We appreciate the interest expressed by Ando et al¹ in our publication describing the development and validation of the RSClin educational tool² and the opportunity to address their comments.

The first concern is related to the estimation of absolute chemotherapy benefit provided by RSClin, a new feature not included in the RSPC model using older trial data.³ Ando et al¹ stated, “The majority of patients enrolled in the two studies [B-20 and TAILORx] received cyclophosphamide, methotrexate, and fluorouracil [CMF], or methotrexate and fluorouracil [MF].,” which could result in underestimation of chemotherapy benefit. Of the 4,921 patients randomly assigned to chemotherapy in TAILORx or B-20 and included in the meta-analysis, only 565 (11%) received CMF/MF. The remaining patients who received chemotherapy had regimens that included anthracyclines, taxanes, or both. In the TAILORx arm including those with a recurrence score (RS) of 26 or higher treated with adjuvant chemotherapy, distant recurrence rates were similar irrespective of whether the patients received regimens including anthracyclines, taxanes, or both.⁴ Although CMF may be less effective in preventing distant recurrence than such regimens,⁵ it may be perfectly appropriate choice for older women or those with comorbidities.

Also relevant to the concern about estimation of chemotherapy benefit, the statement by Ando et al¹ that the RxPONDER trial “…showed no association between the high RS results and the benefit of chemotherapy …” fails to acknowledge that the interaction was only examined for patients with RS 0-25. RxPONDER did not randomly assign patients with a 21-gene RS of 26 or higher who contributed to the significant RS and chemotherapy treatment interaction in B20 and SWOG 8814 and who derive the greatest chemotherapy benefit. Clinical outcomes observed in a population-based cohort of 70,000 women with node-negative breast cancer are consistent with observations from the meta-analysis that greater chemotherapy benefit is observed with a higher 21-gene RS and that the threshold for chemotherapy benefit varies with age (a known surrogate for menopausal status),⁶ a factor now accounted for in the RSClin model. As we demonstrated in our report, the incremental chemotherapy benefit observed with higher 21-gene RS is driven not only by a higher underlying recurrence risk but also by prediction of greater relative chemotherapy benefit with increasing RS.

A second concern raised included the use of age rather than menopausal status as a prognostic covariate and lack of ovarian function suppression as a therapeutic option in the model. In the TAILORx trial, a comparable proportion of randomly assigned patients were premenopausal or 50 years of age or younger (36% v 34%), the majority of premenopausal women received tamoxifen alone as the initial endocrine therapy, and there was an interaction between chemotherapy benefit, RS, and either age or menopausal status.⁷ Moreover, ovarian function suppression was found to have no beneficial effect in patients with node-negative breast cancer in the combined analysis of the SOFT and TEXT trials.⁸ Neither TAILORx nor RxPONDER was designed to determine whether the benefit of chemotherapy in premenopausal women is largely due to premature chemotherapy-induced menopause, which has been hypothesized as contributing to chemotherapy benefit in this setting.⁹ Additional prospective clinical trials may be required to provide the highest level of evidence supporting this hypothesis, although some evidence retrospectively derived from a prospectively conducted trial supports this supposition.¹⁰

A third concern is related to the use of patients receiving contemporary therapies for model development. Evidence indicates improved prognosis for hormone receptor–positive breast cancer after the year 2000.¹¹ Patients in TAILORx were enrolled between 2006 and 2010, and the majority received contemporary therapies including anthracycline and/or taxane-containing chemotherapy and aromatase inhibitors for postmenopausal women, which are currently the most contemporary adjuvant systemic therapies.

A major finding of our analysis is that RSClin provides more prognostic information than clinical-pathologic features or the 21-gene RS used individually, indicating that it provides the most accurate estimate of underlying distant recurrence risk with endocrine therapy alone. Although treatment based on a 21-gene RS test result alone is clear for the majority of patients, some situations are more nuanced. Incorporation of clinicopathologic information can add value in cases where additional clarity is needed by providing individualized estimates of distant recurrence risk and absolute chemotherapy benefit. Examples include women with high clinical risk features and a lower RS, low clinical risk features and a higher RS, and patients 50 years or younger and an RS in the mid-range. We believe that the RSClin educational tool may provide useful information to support shared decision making in such circumstances.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Reply to K. Ando et al

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).