An illustration · invented records · the research is real
What the 21-gene test can and cannot tell us
Question: for a postmenopausal woman with early, hormone-receptor-positive, HER2-negative, node-negative breast cancer, what can the 21-gene recurrence score test (Oncotype DX) tell doctors about whether adding chemotherapy to anti-estrogen therapy is likely to help, and what are the limits of that evidence? Scope and exclusions
Research run: September 8, 2026. Reviewed by Sarah Kiselycznyk: September 10, 2026.
Start with the short version. The detail is here if you want to explore a question further; you do not need to read every paper to prepare for the visit. This page explains the test in general. Mom and her clinicians can discuss whether it would help with her decisions. The brief · Question log
The short version
The test measures 21 genes in a piece of the tumor and gives a score from 0 to 100. For women in Mom’s general situation, two large randomized trials found that most women with scores of 25 or below did as well on anti-estrogen therapy alone as with chemotherapy added. The strongest evidence concerns groups in which adding chemotherapy did not improve outcomes. Evidence about benefit at higher scores is less direct. It is usually run on the tumor removed at surgery, once the lymph node results are known.
Tier 1 — what is established
Each claim links to a source note with the exact words, the identifier, and how it was checked.
- The score predicts recurrence risk. The original validation, in node-negative women on tamoxifen, found ten-year distant recurrence of about 7% in the low-score group and about 31% in the high-score group. Paik 2004
- Low scores (0–10): anti-estrogen therapy alone, very low recurrence. In TAILORx, women with node-negative disease and a score of 0 to 10 had 99.3% freedom from distant recurrence at five years without chemotherapy. Sparano 2015
- Midrange scores (11–25), node-negative: no added benefit observed for women over 50. TAILORx randomized 6,711 women in this range. Overall outcomes were similar with or without chemotherapy. A benefit appeared only in women 50 or younger with scores 16 to 25. Sparano 2018
- One to three positive nodes, scores 0–25: no chemotherapy benefit for postmenopausal women. RxPONDER found five-year outcomes of 91.9% versus 91.3%, endocrine alone versus with chemotherapy, in postmenopausal women. Premenopausal women did benefit. Kalinsky 2021
- Guidelines describe when clinicians may use the test. ASCO’s 2022 guideline lets clinicians use it to guide chemotherapy decisions in postmenopausal women with ER-positive, HER2-negative disease and zero to three positive nodes, and says treatment decisions must also weigh stage, other health conditions, and the patient’s preferences. ASCO 2022 · ASCO 2019
- How the test is done. On tumor tissue preserved in paraffin, in a central laboratory. In the trials, that tissue came from surgery. Sparano 2015
Tier 2 — what is still being studied
- High scores (26 and up). In TAILORx everyone with a high score got chemotherapy, so the trial did not test chemotherapy against none at high scores. Their outcomes were better than expected, which supports but does not prove a benefit. Sparano 2020 · ASCO’s 2019 update treats scores above 30 as candidates for chemotherapy, and 26 to 30 as a judgment call. ASCO 2019
- Does the score add to what size and grade already tell you? Yes, and vice versa: tumor size and grade still carried information in TAILORx. The authors also warn against reading a “no difference” trial too confidently. Sparano 2019
- How good is the test at predicting benefit, as opposed to risk? An independent UK review (2025) rated the evidence for predicting benefit as weaker than the evidence for predicting risk, and mostly limited to this test. Tappenden 2025
- Why do younger women benefit when older women don’t? Possibly because chemotherapy shuts down the ovaries, not because it kills cancer cells directly. A 2026 analysis found a blood marker of ovarian reserve separated premenopausal women who benefited from those who did not. A running trial (OFSET) tests this directly. This matters to Mom only as the reason the age split exists. Kalinsky 2026 · Freeman 2025 · Trials
- What has changed in practice. In the US, chemotherapy use in postmenopausal women with low or intermediate scores fell steadily through 2022. Freeman 2025
- Thinking and memory during treatment. Women in RxPONDER who received chemotherapy reported more cognitive difficulty over three years, in both menopausal groups. Kang 2026
- Can it be run before surgery, on the biopsy? It can; a small study did so to choose pre-surgery treatment. That is a different use from the one discussed for Mom. Bear 2017
Tier 3 — questions the evidence has not settled
- Whether chemotherapy helps postmenopausal women with high scores was not tested by a randomized comparison in the cited trials. Our September 8, 2026 registry searches found no trial addressing this question. Trials
- Late recurrence. In one comparison the score predicted recurrence in the first five years but not in years five to ten. What the score says about the long term is unsettled. Sgroi 2013 · Eleven-year TAILORx follow-up shows the score still tracks breast-cancer deaths, less so overall survival. Yang 2024
- How the score should be combined with tumor size, grade, Ki-67, and other tests is still being worked out; guidelines allow several tests and do not rank them. ASCO 2022
Questions this suggests for September 15
- After surgery, would you order this test for Mom? If not, what would you rely on instead?
- If her score came back in the 11 to 25 range, what would you recommend, and why?
- If it came back above 25, what would you recommend, given that this range was not randomized in the trials?
Limits of this summary
- Sixteen source notes stand behind this page. Eight rest on abstracts only; five on the authors’ manuscripts; two on the publisher’s full text; one on the trial registry. Every quote was matched to its original by exact string comparison.
- Two papers are reproduced in full in this folder (the two whose licenses allow it); each source note says whether its paper is here and, if not, why. Details
- Three sources carry published corrections; those read do not change a finding cited here (each source note says what its correction changes, and one notice remains unread).
- The findings provide background for a conversation with Mom’s clinicians about her care.
Review and sources: How we researched this records the preparation, review scope, and source checks.