RESUMEN
BACKGROUND: With the increasing use of neoadjuvant chemotherapy and minimally invasive ablative therapy in breast cancer, pretreatment assessment of tumour grade on core needle biopsy (CNB) is increasingly needed. However, grading on CNB is possibly less accurate than grading based on the surgical excision specimen. A systematic review and meta-analysis of the literature was conducted to derive a reliable estimate of the agreement in tumour grading between CNB and subsequent surgical excision. METHODS: Following the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) criteria, Embase, PubMed and the Cochrane Library were searched. Pooled proportions of agreement in grading between CNB and the excision specimen, Cohen's κ and percentages of overestimation and underestimation were calculated. Random-effects models were applied because of substantial heterogeneity, assessed by I2 test. Determinants of the level of agreement in grading were explored with meta-regression. RESULTS: Thirty-four articles were included in the systematic review (6029 patients) and 33 in the meta-analysis (4980 patients). Pooled agreement and κ were 71·1 (95 per cent c.i. 68·8 to 73·3) per cent and 0·54 (0·50 to 0·58) respectively. Underestimation and overestimation occurred in 19·1 (17·1 to 21·3) and 9·3 (7·7 to 11·4) per cent respectively. Meta-regression showed associations between agreement of histological type (positive association) and proportion of patients with oestrogen receptor-positive disease (negative association) and grade agreement. CONCLUSION: Grading on CNB corresponds moderately with grading based on excision specimens, with underestimation in about one in five patients. Incorrect CNB tumour grading has limited clinical implications, as multiple factors influence decision-making for adjuvant systemic therapy.