292P - Radiological-histological size correlation in triple-negative breast cancer (TNBC)

Date 29 September 2014
Event ESMO 2014
Session Poster Display session
Topics Breast Cancer, Early Stage
Imaging, Diagnosis and Staging
Pathology/Molecular Biology
Surgery and/or Radiotherapy of Cancer
Presenter Constance Thibault
Citation Annals of Oncology (2014) 25 (suppl_4): iv85-iv109. 10.1093/annonc/mdu327
Authors C. Thibault1, M. Gosset2, F. Chamming'S3, M. Lefrere-Belda4, N. Pecuchet5, L. Fournier3, H. Roussel4, S. Oudard1, F. Lécuru2, J. Medioni1
  • 1Medical Oncology, Hôpital Européen Georges Pompidou, 75015 - Paris/FR
  • 2Gynecologic Surgery, Hôpital Européen Georges Pompidou, Paris/FR
  • 3Radiology Departement, Hôpital Européen Georges Pompidou, Paris/FR
  • 4Pathology Department, Hôpital Européen Georges Pompidou, Paris/FR
  • 5Medical Oncology, Hôpital Européen Georges Pompidou, Paris/FR

Abstract

Aim

Preoperative size of breast tumors is a crucial parameter for guiding modality of surgery and neoadjuvant chemotherapy (NAC). Mammography (MG) and breast ultrasound (US) are known to poorly predict tumor size. MRI is considered to be the best imaging but the low specificity limits its use. Few data are available in the TNBC population.

Methods

We retrospectively collected data on patients (pts) treated for a local TNBC in Georges Pompidou European Hospital (Paris, France) between 2000 and 2012 . To be included, pts had to fulfilled the following inclusion critera: ductal or lobular carcinoma, ER < 10%, PR < 10% and HER-2 negative. Radiological sizes on MG, US and MRI at initial diagnosis were collected from the medical reports when available. The gold standard was postoperative histological size. We considered that tumor was underestimated when variation size (DS) was ≥ - 3 mm, overestimated when DS ≥ + 3 mm and adequately estimated when DS was comprised between – 3 and + 3 mm. We calculated the sensitivity and specificity of each imaging technique to identify ≥ pT2 tumors

Results

Eighty-five tumors were analyzed (Table 1). The correlation coefficient between radiological and pathological size was 0,66 for MG (n = 40), 0,60 for US (n = 60) and 0,74 for MRI (n = 10). MG, US and MRI underestimated tumor size in respectively 58%, 53% and 10%. MG, US and MRI correctly identified ≥ pT2 tumors with a sensitivity and specificity of respectively 65% and 100% (MG), 48% and 92% (US), 100% and 71% (MRI). The frequency of positive resection margin was identical when tumors were underestimated versus [over or correctly] estimated (21% vs 14%, p = 0.38).

Pts characteristics Population n = 85 tumors MG n = 40 US n = 60 MRI n = 10
Age (median, years) 56 (38-88) 58 56 47
cT -T0 - T1-T2 - T3-T4 - unknown 3 (4%) 76 (89%) 5 (6%) 1 (1%) 1(2%) 36 (90%) 3 (8%) 0 1 (2%) 54 (90%) 4 (7%) 1 (2%) 1 (10%) 9 (90%) 0 0
Type of surgery - tumorectomy - mastectomy - unknown 61 (72%) 23 (27%) 1 (1%) 29 (73%) 10 (25%) 1 (2%) 43 (72%) 17 (28%) 0 6 (60%) 4 (40%) 0
Histological type - Ductal - Lobular - Unknown 82 (96%) 2 (3%) 1 (1%) 39 (98%) 0 1 (2%) 58 (%) 2 (%) 0 (0%) 10 (100%) 0 0
pT (median, mm) 18 22 18 12
pN - N+ - N- - Unknown 26 (31%) 58 (68%) 1 (1%) 14 (35%) 25(63%) 1 (2%) 18 (30%) 42 (70%) 0 4 (40%) 6 (60%) 0
SBR - I - II - III - unknown 5 (6%) 29 (34%) 48 (56%) 3 (4%) 0 12 (30%) 26 (65%) 0 5 (8%) 24 (40%) 29 (48%) 2 (3%) 0 3 (30%) 7 (70%) 0
Resection margin - negative - positive - unknown 67 (79%) 17 (20%) 1 (1%) 30 (75%) 10 (25%) 0 53 (88%) 7 (12%) 0 7 (70%) 3 (30%) 0

Conclusions

MG and US tend to underestimate TNBC size without impacting on positivity of resection margin. MG and US discriminated tumors ≥pT2 with a high specificity and therefore adequately exclude patients that do not require NAC. These results are consistent with other data reported in general breast cancer population. MRI better evaluated size in our cohort but the small sample size limits conclusion about the role of MRI in preoperative imaging in TNBC.

Disclosure

All authors have declared no conflicts of interest.