Background Multigene signatures (MGS) select women with estrogen receptor positive human epidermal growth factor receptor 2 negative (ER+/HER2-) breast cancers where adjuvant chemotherapy (aCT) can be avoided. However, MGS are expensive and not always reimbursed. We investigated the predictive value of five inexpensive statistical models in tumors with low or high risk of relapse based on MGS and investigated the change in decision to add chemotherapy following MGS results.

Patients and Methods In this retrospective study, we evaluated patients diagnosed with primary operable ER+/HER2- lymph node negative or positive breast cancer diagnosed at University Hospitals Leuven between 2013 and 2018. Patients were analyzed by MammaPrint® (MP) (n=24), OncotypeDX® (ODX) (n=44) or Prosigna®(n=57) as there was uncertainty about benefit of aCT during multidisciplinary meeting (MDM). Magee equations (ME), Memorial Sloan Kettering simplified score (MSK), Breast Cancer Recurrence Score Estimator (BCRSE), new Adjuvant! Online (nAOL) and PREDICT v2.0 were computed. TAILORx cut-offs were used for ODX. A 5% cut-off was used for 10-year survival benefit with aCT for nAOL and PREDICT.

Results All ME- and BCRSE-high cases were classified by MGS as high or intermediate and not as MGS-low risk, as shown in Table 1. None of the low risk classifications by ME and nAOL resulted in MGS-high risk with ODX. High risk classification with nAOL showed strong concordance with all MGS-high risk results. Chemotherapy switch according to MGS results was observed in 46% (57/125) of patients. Following MGS testing, aCT was given to 56 patients which resulted in 19% relative and 10% absolute reduction.

Conclusion Inexpensive statistical models based on pathologic parameters can be useful to select patients who may need MGS testing. Integration of MGS into MDM decisions, resulted in a substantial decisional switch and reduction in aCT administration.

Table 1

Predictive value of inexpensive statistical models in MGS tested tumors.

  MGS high risk (n=52)   MGS low risk (n=52)  
 ODX (n=17) MP (n=10) Prosigna (n=25) ODX (n=27) MP (n=14) Prosigna (n=11) 
MSK high 59% (10/17) 30% (3/10) 32% (8/25) 4% (1/27) 36% (5/14) 0% (0/11) 
ME high 24% (4/17) 10% (1/10) 4% (1/25) 0% (0/27) 0% (0/14) 0% (0/11) 
BCRSE high 0% (0/17) 10% (1/10) 4% (1/25) 0% (0/27) 0% (0/14) 0% (0/11) 
nAOL high 100% (17/17) 60% (6/10) 96% (24/25) 85% (23/27) 86% (12/14) 27% (3/11) 
PREDICT high 47% (8/17) 40% (4/10) 48% (12/25) 26% (7/27) 36% (5/14) 0% (0/11) 
MSK low 18% (3/17) 30% (3/10) 24% (6/25) 41% (11/27) 29% (4/14) 46% (5/11) 
ME low 0% (0/17) 10% (1/10) 8% (2/25) 7% (2/27) 0% (0/14) 18% (2/11) 
BCRSE low 18% (3/17) 30% (3/10) 40% (10/25) 26% (7/27) 43% (6/14) 64% (7/11) 
nAOL low 0% (0/17) 40% (4/10) 4% (1/25) 15% (4/27) 14% (2/14) 73% (8/11) 
PREDICT low 53% (9/17) 60% (6/10) 52% (13/25) 74% (20/27) 64% (9/14) 100% (11/11) 
  MGS high risk (n=52)   MGS low risk (n=52)  
 ODX (n=17) MP (n=10) Prosigna (n=25) ODX (n=27) MP (n=14) Prosigna (n=11) 
MSK high 59% (10/17) 30% (3/10) 32% (8/25) 4% (1/27) 36% (5/14) 0% (0/11) 
ME high 24% (4/17) 10% (1/10) 4% (1/25) 0% (0/27) 0% (0/14) 0% (0/11) 
BCRSE high 0% (0/17) 10% (1/10) 4% (1/25) 0% (0/27) 0% (0/14) 0% (0/11) 
nAOL high 100% (17/17) 60% (6/10) 96% (24/25) 85% (23/27) 86% (12/14) 27% (3/11) 
PREDICT high 47% (8/17) 40% (4/10) 48% (12/25) 26% (7/27) 36% (5/14) 0% (0/11) 
MSK low 18% (3/17) 30% (3/10) 24% (6/25) 41% (11/27) 29% (4/14) 46% (5/11) 
ME low 0% (0/17) 10% (1/10) 8% (2/25) 7% (2/27) 0% (0/14) 18% (2/11) 
BCRSE low 18% (3/17) 30% (3/10) 40% (10/25) 26% (7/27) 43% (6/14) 64% (7/11) 
nAOL low 0% (0/17) 40% (4/10) 4% (1/25) 15% (4/27) 14% (2/14) 73% (8/11) 
PREDICT low 53% (9/17) 60% (6/10) 52% (13/25) 74% (20/27) 64% (9/14) 100% (11/11) 

Citation Format: Laurence Slembrouck, Giuseppe Floris, Hans Wildiers, Ann Smeets, Erik Van Limbergen, Philippe Moerman, Caroline Weltens, Kevin Punie, Griet Hoste, Els Van Nieuwenhuysen, Sileny Han, Ines Nevelsteen, Lynn Jongen, Patrick Neven, Isabelle Vanden Bempt. Multigene signatures based risk estimates in ER+/HER2- breast cancers: The predictive value of inexpensive statistical models and changes in adjuvant chemotherapy use [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 1406.