Abstract
Background
There are currently no molecular tests to identify individual breast cancers where radiotherapy (RT) offers no benefit. Profile for the Omission of Local Adjuvant Radiotherapy (POLAR) is a 16-gene molecular signature developed to identify low-risk cancers where RT will not further reduce recurrence rates.
Methods
An individual participant data meta-analysis was performed in 623 patients of node-negative estrogen receptor–positive and HER2-negative early breast cancer enrolled in 3 RT randomized trials for whom primary tumor material was available for analysis. A Cox proportional hazards model on time to locoregional recurrence was used to test the interaction between POLAR score and RT.
Results
A total of 429 (69%) patients’ tumors had a high POLAR score, and 194 (31%) had a low score. Patients with high POLAR score had, in the absence of RT, a 10-year cumulative incidence of locoregional recurrence (20%, 95% confidence interval [CI] = 15% to 26%, vs 5%, [CI] 2% to 11%) for those with a low score. Patients with a high POLAR score had a large benefit from RT (hazard ratio [HR] for RT vs no RT = 0.37, 95% CI = 0.23 to 0.60; P < .001). In contrast, there was no evidence of benefit from RT for patients with a low POLAR score (HR = 0.92, 95% CI = 0.42 to 2.02; P = .832). The test for interaction between RT and POLAR was statistically significant (P = .022).
Conclusions
POLAR is not only prognostic for locoregional recurrence but also predictive of benefit from RT in selected patients. Patients aged 50 years and older with estrogen receptor–positive and HER2-negative disease and a low POLAR score could consider omitting adjuvant RT. Further validation in contemporary clinical cohorts is required.
Introduction
The addition of adjuvant radiation therapy (RT) to breast-conserving surgery has consistently demonstrated in randomized controlled trials (RCTs) a reduction in the risk of locoregional recurrence and remains standard of care with surgery and systemic therapy in the management of early breast cancer.1-3 The Early Breast Cancer Trialists’ Collaborative Group demonstrated approximately 65% relative risk reduction in local recurrence at 10 years for patients who received RT after breast-conserving surgery. This translated into a modest absolute improvement in breast cancer survival at 15 years of 3.8%.4,5 The Cancer and Leukemia Group B (CALGB) 9343 and Post-operative Radiotherapy in Minimum-risk Elderly (PRIME) II trials studied RT omission in older breast cancer patients (aged 70 years and older and 65 years and older, respectively) and showed that cancers with low-risk clinicopathologic features have lower rates of local recurrence after breast-conserving surgery and adjuvant endocrine therapy than earlier randomized trials, though local recurrence risks were further reduced by the addition of RT.6,7
Standard clinical and pathologic variables have not identified individual patients with cancers of sufficiently low locoregional recurrence risk with long-term follow-up in whom RT offers no additional tumor control benefit and can be safely omitted.7,8 Despite guidelines supporting the omission of radiation in women aged 70 years and older with T19,10 or some small sized T211 estrogen receptor–positive and HER2-negative tumors treated by breast-conserving surgery and adjuvant endocrine therapy, the use of adjuvant RT in the United States in this setting remains common.12 Identification of biomarkers that predict response of any solid tumor13 or breast cancer in particular to radiation14 is a major research gap.
Current biomarkers to optimize use of chemotherapy in breast cancer prognosticate for distant recurrences.15,16 A recent systematic review of clinical biomarkers of tumor radiosensitivity showed that no gene signature is sufficiently validated for routine use.17 Previous attempts to identify radiation-specific signatures have conflicting results and have often not been breast cancer specific.17-24
An ideal biomarker would not only identify cancers at low risk of local recurrence in the absence of RT (prognostic) but also identify cancers for those whom RT does not meaningfully further reduce that risk and, conversely, for those whom RT would remain beneficial (predictive).14,25,26
Building on our original results,27 we report an individual participant data meta-analysis to validate the Profile for the Omission of Local Adjuvant Radiation (POLAR) biomarker test in patients with low-risk, hormone receptor–positive, HER2-negative, node-negative cancers from 3 RCTs of breast-conserving surgery with or without RT.
Methods
Randomized trials used
Patients from 3 RCTs were included: the Scottish Conservation trial,1 SweBCG91-RT trial,2 and the Princess Margaret RT trial (see Table 1; Tables S1 and S2).3 Complete information on patient eligibility for each trial is in the online supplement (Supplementary Methods). In brief, for the Scottish Conservation trial, eligible patients were aged younger than 70 years with node-negative or node-positive invasive breast cancer that was 4 cm or less in diameter. For SweBCG91-RT, patients were eligible if they were aged younger than 76 years with node-negative, stage I-II invasive breast cancer. For the Princess Margaret trial, eligible patients were aged 50 years or older with node-negative invasive breast cancer that was 5 cm or less in diameter. Information on patient inclusion in the meta-analysis is in Figure S1. The distribution of microarray gene expression in the Princess Margaret samples differed from the other 2 trials and required additional rescaling before analysis (Supplementary Methods, Figure S2, Table S3). Informed consent was obtained from all patients, and approval to conduct the studies was received from the appropriate governing bodies in each country. No data were collected on racial or ethnic groups.
Table 1.
Trial characteristics and patient demographic and clinical characteristics in the individual participant data meta-analysis (n = 623)a
| Characteristics | Scottish Conservation Trial | SweBCG91-RT | Princess Margaret | Overall |
|---|---|---|---|---|
| Trial characteristics | ||||
| No. in original trial | 585 | 1178 | 769 | — |
| Enrollment period | 1985-1991 | 1991-1997 | 1992-2000 | — |
| Stage, tumor size | Tumor < 4 cm | Stage I-IIA | Stage I-II, tumor < 5 cm | — |
| RT administration | 50 Gy, 20-25 fractions | 48-54 Gy, 24-27 fractions | 40 Gy, 16 fractions | — |
| Boost therapy | 10-15 Gy by fractionated external beams or 20-30 Gy by iridium implant | None | 12.5 Gy, 5 fractions | — |
| Systemic therapy | Tamoxifen or chemotherapy per receptor status | Received by 8% of patients | Tamoxifen | — |
| Patient characteristics in individual participant data meta-analysis, No. (%) | ||||
| No. in meta-analysis | 137 | 354 | 132 | 623 |
| Age at surgery, No. (%), y | ||||
| 39 and younger | 5 (3.6) | 7 (2) | 0 | 12 (1.9) |
| 40-49 | 25 (18.2) | 60 (16.9) | 0 | 85 (13.6) |
| 50-59 | 49 (35.8) | 103 (29.1) | 30 (22.7) | 182 (29.2) |
| 60-69 | 58 (42.3) | 120 (33.9) | 47 (35.6) | 225 (36.1) |
| 70-79 | 0 | 64 (18.1) | 46 (34.8) | 110 (17.7) |
| 80 and older | 0 | 0 | 9 (6.8) | 9 (1.4) |
| Histologic grade, No. (%) | ||||
| Grade 1 | 37 (27.0) | 60 (16.9) | 18 (13.6) | 115 (18.5) |
| Grade 2 | 78 (56.9) | 226 (63.8) | 79 (59.8) | 383 (61.5) |
| Grade 3 | 22 (16.1) | 59 (16.7) | 30 (22.7) | 111 (17.8) |
| Missing | 0 | 9 (2.5) | 5 (3.8) | 14 (2.2) |
| Tumor size, No. (%), mm | ||||
| 0-20 | 100 (73.0) | 335 (94.6) | 106 (80.3) | 541 (86.8) |
| 21-40 | 27 (19.7) | 19 (5.4) | 26 (19.7) | 72 (11.6) |
| Missing | 10 (7.3) | 0 | 0 | 10 (1.6) |
| Subtype by immunohistochemistry, No. (%) | ||||
| Luminal A-like | 80 (58.4) | 228 (64.4) | 43 (32.6) | 351 (56.3) |
| Luminal B-like (HER2 negative) | 57 (41.6) | 126 (35.6) | 30 (22.7) | 213 (34.2) |
| Estrogen receptor positive and HER2 negative and not otherwise specified | 0 | 0 | 59 (44.7) | 59 (9.5) |
| Adjuvant systemic therapy, No. (%) | ||||
| Endocrine therapy only | 128 (93.4) | 0 (0) | 132 (100) | 260 (41.7) |
| Chemotherapy only | 9 (6.6) | 0 | 0 | 9 (1.4) |
| None | 0 | 354 (100) | 0 | 354 (56.8) |
| Radiotherapy, No. (%) | ||||
| No | 69 (50.4) | 178 (50.2) | 62 (47.0) | 309 (49.6) |
| Yes | 68 (49.6) | 176 (49.7) | 70 (53.0) | 314 (50.4) |
| Locoregional recurrence, No. (%) | ||||
| No | 109 (79.6) | 282 (79.7) | 116 (87.9) | 507 (81.4) |
| Yes | 28 (20.4) | 72 (20.3) | 16 (12.1) | 116 (18.6) |
| Follow-up time in patients without locoregional recurrence, No. (%), y | ||||
| Median | 21.1 | 13.3 | 8.6 | Not available |
Patients in the 3 trials were randomly assigned to ± whole breast radiotherapy following breast-conserving surgery. Patients in the meta-analysis included the subset from each trial of node-negative patients who had estrogen receptor–positive, HER2-negative tumors and tissue available for analysis.
Development of POLAR
The development of POLAR has been described previously.27 In short, gene expression (Gene Expression Omnibus, GSE119295) from the primary tumors of 764 participants in the SweBCG91-RT was generated from GeneChip Human Exon 1.0 ST Arrays (Thermo Fisher Scientific, South San Francisco, CA, USA) in a Clinical Laboratory Improvement Amendments (CLIA) and College of American Pathologists (CAP)-certified laboratory (Decipher Biosciences, San Diego, CA, United States). The tumors from patients with hormone receptor–positive, HER2-negative tumors who did not receive systemic therapy were selected (n = 597). The material was divided into a training set (n = 243) and a validation set (n = 354). A ranked list of genes highly prognostic for locoregional recurrence in a Cox model were entered into gene set enrichment analysis for patients in the training cohort who did not receive RT (n = 131). The most prognostic genes were further filtered by selecting those in the most enriched biological pathways and with a standard deviation in the highest quartile. The resulting 82 genes were fed into an elastic net regression model in the full training cohort using locoregional recurrence as the endpoint, which resulted in a final model of 16 genes. The score was dichotomized into POLAR high- and low-risk groups using a prespecified cutoff at the 25th percentile (Supplementary Methods).
Individual participant data meta-analysis
For validation of POLAR, gene expression, baseline clinical, and outcomes data from the subset of patients with node-negative, estrogen receptor–positive, HER2-negative tumors (centrally reviewed within each trial) within the 3 RCTs were combined into a single dataset. The primary objective was to determine whether POLAR predicts the magnitude of RT benefit following breast-conserving surgery. Secondary endpoints included evaluation of POLAR as a prognostic factor for locoregional recurrence in the absence of RT and differential effect of RT in patients with low and high POLAR scores. Additionally, to look for potential bias in patient selection in the SweBCG validation cohort and to account for the different gene expression distribution in the samples from the Princess Margaret cohort, we conducted meta-analyses of study summary results that used different combinations of the 3 cohorts (Supplementary Methods).
Statistical analysis
Statistical analyses were prespecified in the statistical analysis plan. The primary endpoint was time to locoregional recurrence, defined as the first occurrence of either ipsilateral breast tumor recurrence or regional recurrence. Time zero was defined as the date of random assignment for the SweBCG91RT and Scottish Conservation trials and date of surgery for the Princess Margaret study. A Cox proportional hazards regression model was fitted to the data with effects for the standardized POLAR score, treatment (RT vs no RT), and the interaction term of the score and treatment (POLAR x RT), stratifying by cohort. The proportional hazards assumption was evaluated using scaled Schoenfeld residuals and did not appear to hold for the model that included the POLAR x RT interaction term (Table S10). Therefore, hazard ratios (HRs) presented for that model should be interpreted as the mean over the follow-up period. Cumulative incidence of locoregional recurrence in each treatment group and POLAR risk group (high vs low) were computed using a competing risk approach (R cmprsk package). Distant metastasis and death without recurrence (but not second cancers) were considered competing events. Patients with locoregional recurrence within 3 months of metastasis were classified as having locoregional recurrence. Univariable Cox proportional hazards regression models were fitted for the POLAR-low and POLAR-high subgroups to assess the association between treatment and locoregional recurrence. Univariable and multivariable Cox proportional hazards regression models assessing the association of the continuous POLAR score with locoregional recurrence were performed in patients not treated with RT. A test for heterogeneity across studies was performed on the interaction log hazard ratio. P values were 2-sided, and those less than .05 were considered statistically significant. Hazard ratios are reported with 95% confidence intervals (CIs).
Results
Data according to the eligibility criteria from the SweBCG91RT validation cohort (n = 354), the Princess Margaret trial (n = 132), and the Scottish Conservation Trial (n = 137) were combined for a total of 623 patients in the individual participant data meta-analysis. Baseline demographics and clinical characteristics are included in Table 1. A total of 194 (31%) patients were categorized into the POLAR low-risk group (Figure 1, A); the remaining 429 patients were POLAR high risk (Figure 1, B). We observed the expected benefit from RT in patients with POLAR high tumors (10-year locoregional recurrence in no RT arm: 20%, 95% CI = 15% to 26%; 10-year locoregional recurrence in RT arm: 7%, 95% CI = 4% to 11%; HR = 0.37, 95% CI = 0.23 to 0.60; P < .001). In contrast, there was no benefit from RT for the POLAR-low tumors (10-year locoregional recurrence in no RT arm: 5%, 95% CI = 2% to 11%; 10-year locoregional recurrence in RT arm: 7%, 95% CI = 3% to 14%; HR = 0.92, 95% CI = 0.42 to 2.02; P = .832).
Figure 1.
Cumulative incidence of locoregional recurrence within 10 years for each treatment arm by POLAR risk group in 194 POLAR low (A) and 429 POLAR high (B). 95% confidence intervals for hazard ratios and cumulative incidences are given in parentheses. CI = confidence interval; Cum. inc. = cumulative incidence; HR = hazard ratio; POLAR = Profile for the Omission of Local Adjuvant Radiation; RT = radiation therapy.
In the meta-analysis, there was a statistically significant interaction between RT and the continuous POLAR score (P = .022), demonstrating the POLAR score was predictive of RT benefit. The results were consistent between the 3 RCTs (Figure 2), and there was no evidence of heterogeneity between studies (Supplementary Methods). The risk of locoregional recurrence at 10 years was calculated as a function of the continuous POLAR score for each treatment arm based on the results from the interaction model (Figure 3).
Figure 2.
Hazard ratios for Profile for the Omission of Local Adjuvant Radiation (POLAR) x radiotherapy interaction. The meta-analysis hazard ratio was statistically significant (P = .022). CI = confidence interval; HR = hazard ratio; n/N = number of locoregional recurrences/number of participants.
Figure 3.
Likelihood of locoregional recurrence at 10 years as a continuous function of the POLAR score percentile for the RT and no RT treatment groups (n = 623). Shaded areas indicate 95% confidence intervals. LRR = locoregional recurrence; POLAR = Profile for the Omission of Local Adjuvant Radiation; RT = radiation therapy
Among patients not receiving RT (n = 309), the continuous POLAR score was prognostic for locoregional recurrence in univariable and multivariable Cox proportional hazards analysis, while tumor size, grade, and immunohistochemistry-based grouping28 was not statistically significant (Table 2). As previously noted,4,5 decreased locoregional recurrence among older women remained statistically significant in our multivariable analysis, with the risk of locoregional recurrence in women aged older than 60 years being notably lower compared with those aged younger than 60 years.
Table 2.
Univariable and multivariable Cox proportional hazards models on time to locoregional recurrence in the no radiation therapy arm (n = 309).
| Univariable |
Multivariable |
|||
|---|---|---|---|---|
| Variable | HR (95% CI) | P | HR (95% CI) | P |
| Profile for the Omission of Local Adjuvant Radiation score (POLAR), continuous, standardized | 1.53 (1.24 to 1.91) | <.001 | 1.43 (1.12 to 1.82) | .005 |
| Age, y | ||||
| Younger than 50 | 1.00 (Referent) | 1.00 (Referent) | ||
| 50-59 | 0.64 (0.37 to 1.13) | .122 | 0.61 (0.33 to 1.14) | .121 |
| 60-69 | 0.53 (0.30 to 0.93) | .028 | 0.45 (0.24 to 0.84) | .012 |
| 70 and older | 0.28 (0.11 to 0.73) | <.001 | 0.17 (0.05 to 0.59) | .005 |
| Tumor size | ||||
| T1 | 1.00 (Referent) | 1.00 (Referent) | ||
| T2 | 1.05 (0.52 to 2.11) | .890 | 1.12 (0.50 to 2.51) | .777 |
| Grade | ||||
| 1 | 1.00 (Referent) | 1.00 (Referent) | ||
| 2 | 1.48 (0.78 to 2.80) | .228 | 1.40 (0.71 to 2.79) | .333 |
| 3 | 1.91 (0.92 to 3.97) | .083 | 1.10 (0.47 to 2.60) | .820 |
| Molecular groupings, approximated by immunohistochemistry | ||||
| Luminal A-like | 1.00 (Referent) | 1.00 (Referent) | ||
| Luminal B-like (HER2 negative) | 1.26 (0.78 to 2.03) | .353 | 1.37 (0.80 to 2.33) | .248 |
CI = confidence interval; HR = hazard ratio.
In a subgroup analysis of patients aged younger than 50 years, there was no evidence of a predictive effect of RT in the dichotomous POLAR risk groups, which was likely because of, at least partially, to the small number (n = 97) of patients in this subgroup (Figure S3). In this age group, patients with POLAR-low tumors and no RT had 10-year locoregional recurrence of 14% (95% CI = 2% to 38%) (Table S6, Figure S4). Two stage, study-level meta-analysis results are in Tables S4 and S5. Results were consistent with the individual patient data meta-analysis presented in this paper and with results previously published that combined SweBCG with Princess Margaret data.27 There is no evidence that POLAR predicts RT benefit or prognosticates for distant metastases or mortality (Figures S5-S10, Tables S7-S9). There was also no evidence of informative censoring (Figures S11 and S12). Distributions of the continuous POLAR result by clinicopathologic factors are in Figure S13.
Discussion
We report, to our knowledge, the first molecular classifier (POLAR), validated in data from 3 randomized RT trials, that is predictive of RT benefit and prognostic for locoregional recurrence based on individual molecular tumor characteristics in patients with estrogen receptor–positive, HER2-negative, node-negative early breast cancer after breast-conserving surgery.
We identified 3 RCTs, which randomly assigned early stage breast cancer patients to RT or not after breast-conserving surgery, with mature long-term follow-up. Analysis of tumor specimens from each of these 3 RCTs demonstrated that POLAR was prognostic for locoregional recurrence in the no RT arms, although this analysis did not reach statistical significance in the smaller Princess Margaret dataset.27 When tumor samples were dichotomized as POLAR high vs POLAR low using the predefined cutoff, all 3 trials demonstrated the expected benefit of RT among women whose tumors were POLAR high but no statistically significant benefit for RT in the POLAR-low group. More importantly, there was a statistically significant interaction between POLAR as a continuous variable and RT in the meta-analysis, which was not identified in the individual trials.
Local breast RT is associated with risks of acute and late toxicity, including cardiac morbidity and mortality,29 as well as second cancers.30 Although toxicity rates after breast RT are low, given the prevalence of early stage breast cancer, the absolute number of patients at risk of toxicity is high, leading to a substantial number of patients whose daily lives could be adversely impacted by their treatments.31 Therefore, the improved selection of low-risk cancers treated by breast-conserving surgery and adjuvant endocrine therapy would benefit patients by avoidance of acute and late toxicities and the health-care system by reducing the workload for RT delivery. Additionally, the financial hardship experienced by patients can be substantial and disproportionately affects women from lower socioeconomic backgrounds and may contribute to further inequities in cancer care.32 Clinicians and patients are faced with the dilemma when considering RT omission in older patients with small low-risk hormone receptor–positive, HER2-negative tumors after breast-conserving surgery as to whether an opportunity to further reduce local recurrence by just over 9% at 10 years7 is being missed. POLAR would have major clinical utility in discriminating among this subgroup those patients where RT benefit was likely or not. Biological markers of low risk in single-arm studies, such as Ki-67 in the LUMINA33 and OncotypeDX in the Individualized Decisions for Endocrine therapy Alone (IDEA)34 trials, have also identified subsets of patients with such a low risk of local recurrence that RT may be omitted. Additional research will be necessary to determine whether POLAR can refine decision making on the use of adjuvant RT in patients whose risk of local recurrence was identified as low based on these biomarkers.
Several signatures describing radioresistance and radiosensitivity in breast cancer have been described in a systematic review17 (Table S11). Some signatures were derived in vitro and subsequently tested in nonrandomized clinical materials with few patients omitting post breast-conserving surgery RT.18,35 Radioresistant properties have been defined in patients with recurrence despite RT and in nonrandomized cohorts.36,37 Some signatures have been developed to define the use of postmastectomy RT.19,38 Other signatures have been developed to prognosticate risk for distant recurrence and to select patients for chemotherapy omission. Some signatures have also shown prognostic capability for local recurrence but have not shown predictive value for RT benefit.38-40 Other endpoints such as overall survival or recurrence-free survival have also been used when defining radiosensitivity.18,22,38,41,42 A signature identifying tumor with a low risk of recurrence after breast-conserving surgery while simultaneously identifying those with no additional benefit of RT, derived and validated exclusively in randomized clinical datasets, has only been shown for the POLAR profile.27
Ongoing clinical trials are assessing whether prognostic and predictive genomic signatures used to determine chemotherapy benefit (eg, OncotypeDX, Mammaprint, ProSigna) might also be able to determine benefit of RT. For example, phase III randomized trials like the NRG BR007 De-escalating Breast Radiation After Lumpectomy for Low Risk, Estrogen Receptor Positive, Breast Cancer (DEBRA) trial34 using the OncotypeDx score and the Examining Personalised Radiation Therapy for Low-Risk Early Breast Cancer (EXPERT) trial using the ProSigna test are examining whether these genomic tests might also identify women who do not benefit from breast RT. Indeed, there is great interest in determining whether these tests might serve a dual purpose and simultaneously identify women who do not benefit from chemotherapy or RT. Previous analyses within the SweBCG91RT trial, however, have suggested that the 21-gene–like and the 70-gene–like signatures may not provide predictive information about RT benefit in women with early stage estrogen receptor–positive breast cancer, with no interaction seen between RT and the signatures, P values of .93 and .51, respectively.21 Thus, there continues to be a need for radiation-specific tests like POLAR to predict RT benefit and to spare patients with radioresistant phenotypes from the risk of cardiac toxicity and radiation-induced second malignancies.29,30 Of note, only 1 of the POLAR genes, Matrix Metallopeptidase 11, overlaps with ProSigna, and no genes overlap with OncotypeDX, suggesting POLAR may reflect different aspects of tumor radiobiology.
Based on the eligibility criteria for the 3 trials in our meta-analysis, we suggest these findings are applicable to patients aged 50 years and older with T1-T2 tumors up to 4 cm in size of any histological grade, which are estrogen receptor and/or progesterone receptor–positive and HER2-negative and node negative. The Scottish Conservation Trial included estrogen receptor–poor and node-positive patients who received chemotherapy. These patients were excluded from the meta-analysis, and our results are not applicable to patients who received chemotherapy. Based on data from breast cancer statistics 2022 and the National Cancer Institute’s Surveillance, Epidemiology, and End Results database, approximately 29% of all US breast cancer patients (approximately 70 000 per year) would thus be eligible for testing with POLAR using these criteria, and up to 30% of eligible patients could consider omitting RT without clinically significant risk of recurrence, based on the proportion of POLAR low risk in this study.43 This represents a meaningful number of low-risk patients for whom POLAR could influence shared decision making in the management of early breast cancer. The adoption of POLAR would support guidelines to limit adjuvant RT to those patients with a higher probability of benefit rather than recommending its use irrespective of absolute benefit.12,44 Further, the omission of RT for approximately 30% of the patients eligible for POLAR potentially may reduce the direct costs for RT of approximately US$21 000 000 per year only in the United States.45 However, because a short course of RT of only 5 fractions46 and partial breast irradiation47 are valid options, some patients with a low POLAR score may prefer these more convenient alternatives. Furthermore, the ongoing Europa Trial48 is randomly assigning patients aged 70 years and older with luminal A–like early breast cancer to exclusive partial breast irradiation or to adjuvant endocrine therapy, testing whether the former might replace 5 years of adjuvant endocrine therapy and inform future patients’ choice of postoperative treatment.
There are some limitations to our study. First, the development of POLAR was conducted in 1 cohort (SweBCG) largely treated without adjuvant systemic therapy, and thus, the generalizability to patients receiving modern adjuvant treatments may be questioned. Conversely, the development in a cohort without systemic treatments may help define a molecular classifier for pure radiosensitivity. In the Princess Margaret and the Scottish Conservation trials, adjuvant endocrine treatments were used, and POLAR was able to discriminate RT benefit in these trials. However, more data are needed in patients treated with adjuvant endocrine therapy. Secondly, the numbers of patients with POLAR low-risk tumors were low in the Princess Margaret and Scottish conservation trials. Even if no clear benefit from RT was observed among these patients, a small possible effect of RT in these patient groups cannot be ruled out. The more pronounced effect of RT among patients with POLAR-high tumors in all 3 RCTs and the statistically significant interaction between POLAR and RT in the meta-analysis indicate a predictive value of POLAR. Third, the proportion of the patients for which tumor tissues could be retrieved and POLAR generated was low in 2 of 3 trials, which raises the possibility of unintentional bias, though all samples included met the prespecified criteria for inclusion in the meta-analysis. Further, the distribution of microarray results from Princess Margaret was subject to rescaling, which was not necessary in the other 2 cohorts. Although the rescaling did not impact analyses using continuous POLAR, it affected POLAR risk group assignment. Additionally, POLAR would require validation across racial and ethnic groups—data we did not collect.
Despite these limitations, this meta-analysis from 3 RCTs clearly demonstrates the clinical potential for POLAR to be used in smaller estrogen receptor–positive node-negative breast cancer patients to identify those women who do not appear to benefit from the use of postoperative adjuvant RT. This underscores the importance of improved understanding of breast cancer biology and the interaction between biology and patient treatment, taking us 1 step closer to a personalized RT strategy for our patients. Although this approach requires further prospective validation, it brings molecular biology–driven targeting of the use of RT a step closer toward clinical practice.
In conclusion, to our knowledge, POLAR is the first molecular classifier prognostic for locoregional recurrence and predictive of RT benefit. This classifier is an important step toward molecularly stratified targeting of the use of RT.
Supplementary Material
Acknowledgments
The academic and governmental funders had no role or influence in the design of the study, the analyses, or the decision to publish. PFS Genomics and Exact Sciences Corporation performed the work to electronically merge microarray data with clinical data from each of the 3 trials and assisted with the analysis, but the academic researchers did the final interpretation of the data. PFS Genomics and Exact Sciences Corporation had no role in the decision to publish.
Presented in part at the San Antonio Breast Cancer Symposium, December 9, 2022, San Antonio, TX, USA.
Contributor Information
Per Karlsson, Department of Oncology, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, 413 45 Gothenburg, Sweden; Sahlgrenska Comprehensive Cancer Center, Sahlgrenska University Hospital, Gothenburg, Sweden.
Anthony Fyles, Princess Margaret Cancer Centre, University of Toronto, Toronto, ON, Canada.
S Laura Chang, Exact Sciences Corporation, Madison, WI, United States.
Bradley Arrick, Exact Sciences Corporation, Madison, WI, United States.
Frederick L Baehner, Exact Sciences Corporation, Madison, WI, United States.
Per Malmström, Division of Oncology, Department of Clinical Sciences, Lund University, Lund, Sweden; Department of Haematology, Oncology and Radiation Physics, Skåne University Hospital, Lund, Sweden.
Mårtin Fernö, Division of Oncology, Department of Clinical Sciences, Lund University, Lund, Sweden.
Erik Holmberg, Department of Oncology, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, 413 45 Gothenburg, Sweden.
Martin Sjöström, Division of Oncology, Department of Clinical Sciences, Lund University, Lund, Sweden; Department of Radiation Oncology, University of California San Francisco, San Francisco, CA, United States.
Fei-Fei Liu, Princess Margaret Cancer Centre, University of Toronto, Toronto, ON, Canada.
David A Cameron, Cancer Research UK Scotland Centre, Institute of Genetics and Cancer, University of Edinburgh, Edinburgh, United Kingdom; Usher Institute, University of Edinburgh, Edinburgh, United Kingdom.
Linda J Williams, Usher Institute, University of Edinburgh, Edinburgh, United Kingdom.
John M S Bartlett, Cancer Research UK Scotland Centre, Institute of Genetics and Cancer, University of Edinburgh, Edinburgh, United Kingdom.
Joanna Dunlop, Public Health Scotland, Edinburgh, United Kingdom.
Jacqueline Caldwell, Public Health Scotland, Edinburgh, United Kingdom.
Joseph F Loane, Queen Elizabeth University Hospital, Glasgow, Scotland.
Elizabeth Mallon, Queen Elizabeth University Hospital, Glasgow, Scotland.
Tammy Piper, Cancer Research UK Scotland Centre, Institute of Genetics and Cancer, University of Edinburgh, Edinburgh, United Kingdom.
Ian Kunkler, Cancer Research UK Scotland Centre, Institute of Genetics and Cancer, University of Edinburgh, Edinburgh, United Kingdom.
Felix Y Feng, Department of Radiation Oncology, University of California San Francisco, San Francisco, CA, United States.
Corey W Speers, Department of Radiation Oncology, Case Comprehensive Cancer Center, OH, United States.
Lori J Pierce, Department of Radiation Oncology, Rogel Comprehensive Cancer Center, University of Michigan, Ann Arbor, MI, United States.
John P Bennett, Exact Sciences Corporation, Madison, WI, United States.
Karen J Taylor, Cancer Research UK Scotland Centre, Institute of Genetics and Cancer, University of Edinburgh, Edinburgh, United Kingdom.
Author contributions
Per Karlsson, MD, PhD (Conceptualization; Formal analysis; Investigation; Methodology; Writing—original draft; Writing—review & editing), Lori J. Pierce, MD (Conceptualization; Investigation; Writing—review & editing), Corey W. Speers, MD, PhD (Conceptualization; Investigation; Writing—original draft; Writing—review & editing), Felix Y. Feng, MD (Conceptualization; Investigation; Writing—review & editing), Ian Kunkler, FRCPE (Conceptualization; Investigation; Writing—review & editing), Tammy Piper, MSc (Writing—review & editing), Elizabeth Mallon, FRCPath (Writing—review & editing), Joseph F. Loane, FRCPath (Writing—review & editing), Jacqueline Caldwell, MBA (Writing—review & editing), Joanna Dunlop, PhD (Writing—review & editing), John M.S. Bartlett, PhD (Writing—review & editing), Linda J. Williams, PhD (Formal analysis; Methodology; Writing—review & editing), David A. Cameron, MD (Investigation; Writing—review & editing), Fei-Fei Liu, MD, FRCPC (Investigation; Writing—review & editing), Martin Sjöström, MD, PhD (Conceptualization; Formal analysis; Methodology; Writing—review & editing), Erik Holmberg, PhD (Writing—review & editing), Mårtin Fernö, PhD (Writing—review & editing), Per Malmström, MD, PhD (Writing—review & editing), Frederick L Baehner, MD (Writing—review & editing), Bradley Arrick, MD, PhD (Conceptualization; Investigation; Methodology; Writing—review & editing), S. Laura Chang, PhD (Conceptualization; Formal analysis; Investigation; Methodology; Writing—review & editing), Anthony Fyles, MD, FRCPC (Conceptualization; Investigation; Writing—review & editing), John Bennett, MPH (Formal analysis; Methodology; Writing—original draft; Writing—review & editing), and Karen J. Taylor, PhD (Investigation; Writing—review & editing).
Supplementary material
Supplementary material is available at JNCI: Journal of the National Cancer Institute online.
Funding
This work was supported by the Breast Cancer Institute Fund (Edinburgh and Lothian Health Foundation), Canadian Institutes of Health Research, Exact Sciences Corporation, PFS Genomics, Swedish Cancer Society, and Swedish Research Council.
Conflicts of interest
| Author | Disclosures |
|---|---|
| Per Karlsson |
|
| Anthony Fyles |
|
| S. Laura Chang |
|
| Bradley Arrick |
|
| Frederick Baehner |
|
| Per Malmström |
|
| Mårten Fernö |
|
| Erik Holmberg |
|
| Martin Sjöström |
|
| Fei-Fei Liu |
|
| David A. Cameron |
|
| Linda J. Williams |
|
| John MS Bartlett |
|
| Joanna Dunlop |
|
| Jacqueline Caldwell |
|
| Joseph F. Loane |
|
| Elizabeth Mallon |
|
| Tammy Piper |
|
| Ian Kunkler |
|
| Felix Y. Feng |
|
| Corey W. Speers |
|
| Lori J. Pierce |
|
| John Bennett |
|
| Karen J. Taylor |
|
Data availability
Gene expression data from the SweBCG91-RT cohort are available at Gene Expression Omnibus (GEO, accession number GSE119295), and gene expression data from the Princess Margaret cohort are available as a Data Supplement. Additional data are available from the corresponding author upon reasonable request and after additional ethical approval as necessary.
References
- 1. Forrest AP, Stewart HJ, Everington D, et al. Randomised controlled trial of conservation therapy for breast cancer: 6-year analysis of the Scottish trial. Scottish Cancer Trials Breast Group. Lancet. 1996;348:708-713. 10.1016/s0140-6736(96)02133-2 [DOI] [PubMed] [Google Scholar]
- 2. Malmström P, Holmberg L, Anderson H, et al. ; Swedisj Breast Cancer Group. Breast conservation surgery, with and without radiotherapy, in women with lymph node-negative breast cancer: a randomised clinical trial in a population with access to public mammography screening. Eur J Cancer. 2003;39:1690-1697. 10.1016/s0959-8049(03)00324-1 [DOI] [PubMed] [Google Scholar]
- 3. Fyles AW, McCready DR, Manchul LA, et al. Tamoxifen with or without breast irradiation in women 50 years of age or older with early breast cancer. N Engl J Med. 2004;351:963-970. 10.1056/NEJMoa040595 [DOI] [PubMed] [Google Scholar]
- 4. Clarke M, Collins R, Darby S, et al. ; Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Effects of radiotherapy and of differences in the extent of surgery for early breast cancer on local recurrence and 15-year survival: an overview of the randomised trials. Lancet. 2005;366:2087-2106. 10.1016/S0140-6736(05)67887-7 [DOI] [PubMed] [Google Scholar]
- 5. Darby S, McGale P, Correa C, et al. ; Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Effect of radiotherapy after breast-conserving surgery on 10-year recurrence and 15-year breast cancer death: meta-analysis of individual patient data for 10,801 women in 17 randomised trials. Lancet. 2011;378:1707-1716. 10.1016/S0140-6736(11)61629-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Hughes KS, Schnaper LA, Bellon JR, et al. Lumpectomy plus tamoxifen with or without irradiation in women age 70 years or older with early breast cancer: long-term follow-up of CALGB 9343. J Clin Oncol. 2013;31:2382-2387. 10.1200/JCO.2012.45.2615 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Kunkler IH, Williams LJ, Jack WJL, Cameron DA, Dixon JM.. Breast-conserving surgery with or without irradiation in early breast cancer. N Engl J Med. 2023;388:585-594. 10.1056/NEJMoa2207586 [DOI] [PubMed] [Google Scholar]
- 8. Killander F, Karlsson P, Anderson H, et al. No breast cancer subgroup can be spared postoperative radiotherapy after breast-conserving surgery. Fifteen-year results from the Swedish Breast Cancer Group randomised trial, SweBCG 91 RT. Eur J Cancer. 2016;67:57-65. 10.1016/j.ejca.2016.08.001 [DOI] [PubMed] [Google Scholar]
- 9. Carlson RW. The NCCN 22nd annual conference: discussing treatment disparities, the doctor-patient relationship, the latest NCCN guidelines updates, and more. J Natl Compr Canc Netw. 2017;15:653-654. 10.6004/jnccn.2017.0067 [DOI] [PubMed] [Google Scholar]
- 10. Gradishar WJ, Anderson BO, Balassanian R, et al. NCCN guidelines insights: breast cancer, version 1.2017. J Natl Compr Canc Netw. 2017;15:433-451. 10.6004/jnccn.2017.0044 [DOI] [PubMed] [Google Scholar]
- 11. Thomssen C, Balic M, Harbeck N, Gnant M.. St Gallen/Vienna 2021: a brief summary of the consensus discussion on customizing therapies for women with early breast cancer. Breast Care (Basel). 2021;16:135-143. 10.1159/000516114 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Downs-Canner S, Zabor EC, Wind T, et al. Radiation therapy after breast-conserving surgery in women 70 years of age and older: how wisely do we choose? Ann Surg Oncol. 2019;26:969-975. 10.1245/s10434-018-07151-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Forker LJ, Choudhury A, Kiltie AE.. Biomarkers of tumour radiosensitivity and predicting benefit from radiotherapy. Clin Oncol (R Coll Radiol). 2015;27:561-569. 10.1016/j.clon.2015.06.002 [DOI] [PubMed] [Google Scholar]
- 14. Eccles SA, Aboagye EO, Ali S, et al. Critical research gaps and translational priorities for the successful prevention and treatment of breast cancer. Breast Cancer Res. 2013;15:R92. 10.1186/bcr3493 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Sparano JA, Gray RJ, Ravdin PM, et al. Clinical and genomic risk to guide the use of adjuvant therapy for breast cancer. N Engl J Med. 2019;380:2395-2405. 10.1056/NEJMoa1904819 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Piccart M, van ‘t Veer LJ, Poncet C, et al. 70-gene signature as an aid for treatment decisions in early breast cancer: updated results of the phase 3 randomised MINDACT trial with an exploratory analysis by age. Lancet Oncol. 2021;22:476-488. 10.1016/S1470-2045(21)00007-3 [DOI] [PubMed] [Google Scholar]
- 17. Bleaney CW, Abdelaal H, Reardon M, et al. Clinical biomarkers of tumour radiosensitivity and predicting benefit from radiotherapy: a systematic review. Cancers (Basel). 2024;16:1942., 10.3390/cancers16101942 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Eschrich SA, Fulp WJ, Pawitan Y, et al. Validation of a radiosensitivity molecular signature in breast cancer. Clin Cancer Res. 2012;18:5134-5143. 10.1158/1078-0432.CCR-12-0891 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Tramm T, Mohammed H, Myhre S, et al. Development and validation of a gene profile predicting benefit of postmastectomy radiotherapy in patients with high-risk breast cancer: a study of gene expression in the DBCG82bc cohort. Clin Cancer Res. 2014;20:5272-5280. 10.1158/1078-0432.CCR-14-0458 [DOI] [PubMed] [Google Scholar]
- 20. Speers C, Zhao S, Liu M, Bartelink H, Pierce LJ, Feng FY.. Development and validation of a novel radiosensitivity signature in human breast cancer. Clin Cancer Res. 2015;21:3667-3677. 10.1158/1078-0432.CCR-14-2898 [DOI] [PubMed] [Google Scholar]
- 21. Sjöström M, Chang SL, Fishbane N, et al. Clinicogenomic radiotherapy classifier predicting the need for intensified locoregional treatment after breast-conserving surgery for early-stage breast cancer. Journal of Clinical Oncology: Official Journal of the American Society of Clinical Oncology. 2019;37:s. 3340-3349. In en). 10.1200/JCO.19.00761 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Cui Y, Li B, Pollom EL, Horst KC, Li R.. Integrating radiosensitivity and immune gene signatures for predicting benefit of radiotherapy in breast cancer. Clin Cancer Res. 2018;24:4754-4762. 10.1158/1078-0432.CCR-18-0825 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Torres-Roca JF, Eschrich S, Zhao H, et al. Prediction of radiation sensitivity using a gene expression classifier. Cancer Res. 2005;65:7169-7176. 10.1158/0008-5472.CAN-05-0656 [DOI] [PubMed] [Google Scholar]
- 24. Servant N, Bollet MA, Halfwerk H, et al. Search for a gene expression signature of breast cancer local recurrence in young women. Clin Cancer Res. 2012;18:1704-1715. 10.1158/1078-0432.CCR-11-1954 [DOI] [PubMed] [Google Scholar]
- 25. Woodward WA, Mitchell MP.. The elusive prize of radiation therapy predictive assays in breast cancer. Int J Radiat Oncol Biol Phys. 2023;115:103-105. 10.1016/j.ijrobp.2022.07.2311 [DOI] [PubMed] [Google Scholar]
- 26. Hayes DF. Biomarker validation and testing. Mol Oncol. 2015;9:960-966. 10.1016/j.molonc.2014.10.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Sjöström M, Fyles A, Liu FF, et al. Development and validation of a genomic profile for the omission of local adjuvant radiation in breast cancer. J Clin Oncol. 2023;41:1533-1540. 10.1200/JCO.22.00655 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Sjöström M, Lundstedt D, Hartman L, et al. Response to radiotherapy after breast-conserving surgery in different breast cancer subtypes in the Swedish Breast Cancer Group 91 radiotherapy randomized clinical trial. J Clin Oncol. 2017;35:3222-3229. 10.1200/JCO.2017.72.7263 [DOI] [PubMed] [Google Scholar]
- 29. Darby SC, Ewertz M, McGale P, et al. Risk of ischemic heart disease in women after radiotherapy for breast cancer. N Engl J Med. 2013;368:987-998. 10.1056/NEJMoa1209825 [DOI] [PubMed] [Google Scholar]
- 30. Grantzau T, Mellemkjær L, Overgaard J.. Second primary cancers after adjuvant radiotherapy in early breast cancer patients: a national population based study under the Danish Breast Cancer Cooperative Group (DBCG). Radiother Oncol. 2013;106:42-49. 10.1016/j.radonc.2013.01.002 [DOI] [PubMed] [Google Scholar]
- 31. Lundstedt D, Gustafsson M, Malmstrom P, et al. Symptoms 10-17 years after breast cancer radiotherapy data from the randomised SWEBCG91-RT trial. Radiother Oncol. 2010;97:281-287. 10.1016/j.radonc.2010.09.018 [DOI] [PubMed] [Google Scholar]
- 32. Ehsan AN, Wu CA, Minasian A, et al. Financial toxicity among patients with breast cancer worldwide: a systematic review and meta-analysis. JAMA Netw Open. 2023;6:e2255388. 10.1001/jamanetworkopen.2022.55388 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Whelan TJ, Smith S, Parpia S, et al. ; LUMINA Study Investigators. Omitting radiotherapy after breast-conserving surgery in luminal a breast cancer. N Engl J Med. 2023;389:612-619. 10.1056/NEJMoa2302344 [DOI] [PubMed] [Google Scholar]
- 34. Jagsi R, Griffith KA, Harris EE, et al. Omission of radiotherapy after breast-conserving surgery for women with breast cancer with low clinical and genomic risk: 5-year outcomes of IDEA. J Clin Oncol. 2024;42:390-398. 10.1200/JCO.23.02270 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Weichselbaum RR, Ishwaran H, Yoon T, et al. An interferon-related gene signature for DNA damage resistance is a predictive marker for chemotherapy and radiation for breast cancer. Proc Natl Acad Sci USA. 2008;105:18490-18495. 10.1073/pnas.0809242105 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Wushou A, Jiang YZ, Hou J, Liu YR, Guo XM, Shao ZM.. Development of triple-negative breast cancer radiosensitive gene signature and validation based on transcriptome analysis. Breast Cancer Res Treat. 2015;154:57-62. 10.1007/s10549-015-3611-0 [DOI] [PubMed] [Google Scholar]
- 37. Sjostrom M, Staaf J, Eden P, et al. Identification and validation of single-sample breast cancer radiosensitivity gene expression predictors. Breast Cancer Res. 2018;20:64. 10.1186/s13058-018-0978-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Goodman CR, Seagle BL, Kocherginsky M, Donnelly ED, Shahabi S, Strauss JB.. 21-Gene Recurrence Score Assay Predicts Benefit of Post-Mastectomy Radiotherapy in T1-2 N1 Breast Cancer. Clin Cancer Res. 2018;24:3878-3887. 10.1158/1078-0432.CCR-17-3169 [DOI] [PubMed] [Google Scholar]
- 39. Drukker CA, Elias SG, Nijenhuis MV, et al. Gene expression profiling to predict the risk of locoregional recurrence in breast cancer: A pooled analysis. Breast Cancer Res Treat. 2014;148:599-613. 10.1007/s10549-014-3188-z [DOI] [PubMed] [Google Scholar]
- 40. Fitzal F, Filipits M, Fesl C, et al. ; Austrian Breast and Colorectal Cancer Study Group (ABCSG). PAM-50 predicts local recurrence after breast cancer surgery in postmenopausal patients with ER+/HER2- disease: Results from 1204 patients in the randomized ABCSG-8 trial. Br J Surg. 2021;108:308-314. 10.1093/bjs/znaa089 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Ji Y, Jiang Q, Jian G, et al. Developing a radiosensitivity gene signature for Caucasian patients with breast cancer. Oncol Rep. 2018;40:1695-1705. 10.3892/or.2018.6567 [DOI] [PubMed] [Google Scholar]
- 42. Yan D, Shen M, Du Z, et al. Developing ZNF gene signatures predicting radiosensitivity of patients with breast cancer. J Oncol. 2021;2021:9255494. 10.1155/2021/9255494 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Giaquinto AN, Sung H, Miller KD, et al. Breast Cancer Statistics, 2022. CA Cancer J Clin. 2022;72:524-541. 10.3322/caac.21754 [DOI] [PubMed] [Google Scholar]
- 44. Biganzoli L, Battisti NML, Wildiers H, et al. Updated recommendations regarding the management of older patients with breast cancer: a joint paper from the European Society of Breast Cancer Specialists (EUSOMA) and the International Society of Geriatric Oncology (SIOG). Lancet Oncol. 2021;22:e327-e340. 10.1016/S1470-2045(20)30741-5 [DOI] [PubMed] [Google Scholar]
- 45. Niska JR, Keole SR, Pockaj BA, et al. Choosing wisely after publication of level I evidence in breast cancer radiotherapy. Breast Cancer (Dove Med Press). 2018;10:31-37. 10.2147/BCTT.S153117 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Murray Brunt A, Haviland JS, Wheatley DA, et al. ; FAST-Forward Trial Management Group. Hypofractionated breast radiotherapy for 1 week versus 3 weeks (FAST-Forward): 5-year efficacy and late normal tissue effects results from a multicentre, non-inferiority, randomised, phase 3 trial. Lancet. 2020;395:1613-1626. 10.1016/S0140-6736(20)30932-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Coles CE, Griffin CL, Kirby AM, et al. ; IMPORT Trialists. Partial-breast radiotherapy after breast conservation surgery for patients with early breast cancer (UK IMPORT LOW trial): 5-year results from a multicentre, randomised, controlled, phase 3, non-inferiority trial. Lancet. 2017;390:1048-1060. 10.1016/S0140-6736(17)31145-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Meattini I, Poortmans PMP, Marrazzo L, et al. Exclusive endocrine therapy or partial breast irradiation for women aged >/=70 years with luminal A-like early stage breast cancer (NCT04134598 - EUROPA): Proof of concept of a randomized controlled trial comparing health related quality of life by patient reported outcome measures. J Geriatr Oncol. 2021;12:182-189. 10.1016/j.jgo.2020.07.013 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Gene expression data from the SweBCG91-RT cohort are available at Gene Expression Omnibus (GEO, accession number GSE119295), and gene expression data from the Princess Margaret cohort are available as a Data Supplement. Additional data are available from the corresponding author upon reasonable request and after additional ethical approval as necessary.