Original StudyOncotype DX Predictive Nomogram for Recurrence Score Output: The Novel System ADAPTED01 Based on Quantitative Immunochemistry Analysis
Introduction
Breast cancer is a heterogeneous disease, and the choice of adjuvant treatments is based on specific recurrence risk of relapse due to phenotypical and pathologic tumor characteristics.1, 2, 3 Luminal subtype (estrogen receptor [ER] positive, human epidermal growth factor receptor 2 [HER2] negative) presents itself a variable risk of relapse, so it is difficult for clinicians to escalate or de-escalate therapies in these patients. In clinical practice, most common biomarkers used to define treatments are currently clinicopathologic and immunochemistry features.4 However, while using these tumor characteristics, it is not possible to address patients to optimized, tailored therapies.
Since the 2000s, microarray DNA analysis has improved our knowledge of breast cancer, identifying subtypes corresponding to different clinical outcomes.2,3 On the basis of microarray analysis, further simplified genomic studies permitted instruments to be developed and validated to support clinical decision making.2,3,5, 6, 7 Further analyses developed simplified tests that were based on fewer gene parameters for clinical use. Oncotype DX (ODX) is based on a system of 21-gene reverse transcription (RT) PCR analysis and is correlated retrospectively with prognosis; recent prospective randomized studies have been performed on a node-negative ER+HER2− series of patients.7, 8, 9, 10 Thanks to ODX, adjuvant systemic therapies for early breast cancer have been implemented as a result of identification of patients at a higher risk of cancer who undergo chemotherapy and endocrine therapy, while patients without a significative change in prognosis are not addressed to chemotherapy, thus reducing health costs and sparing patients adverse events.
To date, ODX is the only test included in the definition of adjuvant treatments in the criteria used for classifying risk of relapse (tumor, node, metastasis classification system; American Joint Committee on Cancer staging). In the National Comprehensive Cancer Network guidelines, ODX is preferred for prediction and prognosis, with a category of evidence and consensus of 1.11 In many studies of cost-effectiveness, ODX has been deemed advantageous with regard to standard care as a result of gain in terms of quality-adjusted life-year per year.12,13 However, in many health systems, patients themselves pay for ODX. Of course sometimes patients cannot bear this cost, so they must undergo chemotherapy with no further genotypical data available.
In this study, we retrospectively collected clinicopathologic and immunohistochemical (IHC) features of breast cancer of patients who underwent the ODX test. From these data, which are commonly available in clinical practice, we developed a machine learning model to predict the recurrence score (RS) of ODX. This model led us to create a decision supporting system (DSS), which gives as output a prediction of RS ≤ 25, which in turn addresses requests for the ODX test for cost optimization in systems that cannot afford this health care spending. We called the DSS we created ADAPTED01, and it is freely available online (https://kbolab-tools.shinyapps.io/ADAPTED01/).
Section snippets
Patient Selection
In our analysis, we included all patients who underwent ODX in Fondazione Policlinico Agostino Gemelli IRCCS after surgery for invasive breast cancer. Inclusion criteria were multidisciplinary team indication for the test, ER+HER2− subtype, invasive breast cancer, stage pT1/T2 with pN0-1 disease, and no neoadjuvant systemic therapies provided before surgery. Retrospectively collected clinical data were sex, age at diagnosis, fertility status, clinical staging, and surgery procedure description.
Epidemiologic Results
For the internal data set, 407 patients were included and underwent ODX analysis. Mean (range) age was 53.7 (31-80) years, and 222 patients (54.55%) were > 50 years old. At diagnosis, patients’ menopausal status was as follows: 160 premenopausal (39.3%), 44 perimenopausal (10.8%), 199 postmenopausal (48.9%), and 4 not available—male patients (1%). Tumor characteristics of the internal set are provided in Table 1. ODX results showed the following: 68 patients (16.7%) had scores between 0 and 10,
Rationale and ADAPTED01 Development
ODX is an RT-PCR DNA genomic test that has permitted implementing breast cancer adjuvant treatments, thereby avoiding chemotherapy and its adverse effects, to patients with a low RS.7,19 In 2018, Sparano et al9 published the 10-year results of the TailorX trial, showing that a midrange RS of 11 to 25 presents the same efficacy from adjuvant endocrine therapy and chemoendocrine therapy (hazard ratio, 1.08; 95% confidence interval, 0.94-1.25; P = .26) for women aged > 50 years. The patients in
Conclusion
Quantitative IHC presented a good correlation with RS score in patients with RS ≤ 25 in both the internal and external validation sets. A TRIPOD 3 nomogram for physicians that enhances good and cost-effective clinical practice has been developed with a high positive predictive value that was confirmed in an external validation set, with possibility of 9% to have false positive. Prospective validation of this nomogram in clinical practice could allow resource savings by avoiding unnecessary
Disclosure
The authors have stated that they have no conflict of interest.
References (37)
- et al.
Tailoring therapies—improving the management of early breast cancer: St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer, 2015
Ann Oncol
(2015) - et al.
Adjuvant chemotherapy and timing of tamoxifen in postmenopausal patients with endocrine-responsive, node-positive breast cancer: a phase 3, open-label, randomised controlled trial
Lancet
(2009) - et al.
Cancer Care Ontario guideline recommendations for hormone receptor testing in breast cancer
Clin Oncol
(2012) - et al.
A new framework to enhance the interpretation of external validation studies of clinical prediction models
J Clin Epidemiol
(2015) - et al.
Disparities in compliance with the Oncotype DX breast cancer test in the United States: a National Cancer Data Base assessment
Am J Surg
(2018) - et al.
Budget impact analysis of gene expression tests to aid therapy decisions for breast cancer patients in Germany
Breast
(2018) - et al.
A predictive model for high/low risk group according to Oncotype DX recurrence score using machine learning
Eur J Surg Oncol
(2019) - et al.
Prediction of Oncotype DX recurrence score using clinical parameters: a comparison of available tools and a simple predictor based on grade and progesterone receptor
Hematol Oncol Stem Cell Ther
(2019) Receiver operating characteristic curve in diagnostic test assessment
J Thorac Oncol
(2010)- et al.
Progress in the clinical detection of heterogeneity in breast cancer
Cancer Med
(2016)
Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications
Proc Natl Acad Sci U S A
Molecular portraits of human breast tumours
Nature
A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer
N Engl J Med
A population-based study of tumor gene expression and risk of breast cancer death among lymph node–negative patients
Breast Cancer Res
Prediction of risk of distant recurrence using the 21-gene recurrence score in node-negative and node-positive postmenopausal patients with breast cancer treated with anastrozole or tamoxifen: a TransATAC study
J Clin Oncol
What should physicians look for in evaluating prognostic gene-expression signatures?
Nat Rev Clin Oncol
Adjuvant chemotherapy guided by a 21-gene expression assay in breast cancer
N Engl J Med
Cited by (3)
Shear-wave elastography-based nomograms predicting 21-gene recurrence score for adjuvant chemotherapy decisions in patients with breast cancer
2023, European Journal of RadiologyCitation Excerpt :Histologic grade and tumor subtype in nomograms for RS ≥ 16 and PR status and Ki-67 for RS ≥ 26 were common, regardless of SWE parameters. This finding agrees with previously reported nomograms for RS showing histologic grade, PR status, and Ki-67 as the most common predictors [24,26–30]. Regarding the performance of nomograms, all nomograms in our study except for the Eratio model for RS ≥ 16 were well calibrated and discriminative with AUROC of 0.74 for RS ≥ 16 in two models and 0.81 to 0.86 for RS ≥ 26 in three models [31].
Use of a supervised machine learning model to predict Oncotype DX risk category in node-positive patients older than 50 years of age
2022, Breast Cancer Research and Treatment