Xpert Breast Cancer STRAT4 Assay using fine-needle aspiration biopsy samples in a resource-constrained setting: a prospective diagnostic accuracy study
Dianna L Ng,Edda Vuhahula,Asteria H Kimambo,Marie Claire Ndayisaba,Godfrey S Philipo,Beatrice P Mushi,Kenneth E Ho,Alan Paciorek,Zainab Illonga,Li Zhang,Poonam Vohra,Jodi Weidler,Michael Bates,Elia J Mmbaga,Katherine Van Loon
DOI: https://doi.org/10.1016/S1470-2045(24)00456-X
Abstract:Background: Use of fine-needle aspiration biopsy (FNAB) specimens on Xpert Breast Cancer STRAT4 Assay (STRAT4; Cepheid, Sunnyvale, CA, USA), a CE-marked in-vitro diagnostic medical device, could potentially increase access to breast cancer biomarker testing in resource-constrained settings. We aimed to assess the performance of a research use-only version of STRAT4 using FNAB specimens in Tanzania. Methods: In this prospective diagnostic accuracy study, patients aged 18 years or older with palpable breast masses presenting to the FNAB Clinic at Muhimbili National Hospital (Dar es Salaam, Tanzania) were recruited consecutively. Patients who were pregnant, lactating, or had a previous diagnosis of breast cancer were excluded. STRAT4 testing was performed on off-label FNAB samples using four protocols: the 1 × protocol (using the standard lysate method) on FNAB smears (1 × FNAB), quick lysis and Maui protocols (both on FNAB smears), and the 1 × protocol on formalin-fixed paraffin-embedded (FFPE) cell block material (1 × cell block). For 1 × FNAB and 1 × cell block, tissue was processed using FFPE lysis reagent, incubated at 80°C with proteinase K, and followed by addition of 95% or higher ethanol. Quick lysis was processed using FFPE lysis reagent and 95% or higher ethanol, whereas Maui was processed using a proprietary research-use only lysis reagent. The primary outcomes were overall concordance, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) of STRAT4 as compared with immunohistochemistry or immunohistochemistry plus fluorescence in-situ hybridisation performed on cell blocks using clinically validated protocols in a Clinical Laboratory Improvement Amendments-accredited laboratory at the University of California, San Francisco (San Francisco, CA, USA). Findings: Between Nov 29, 2017, and Dec 17, 2020, 208 patients were enrolled. Of 208 cases, 51 (25%) were excluded from analysis because of insufficient tissue in the cell block or absent cell blocks, leaving 157 participants (all female) for analysis. For oestrogen receptor, 1 × FNAB had the best performance, with an overall concordance of 95% (95% CI 90-100), sensitivity of 94% (85-100), specificity of 97% (90-100), and AUC of 0·96 (0·81-1·00). For progesterone receptor, 1 × cell block had the best overall performance (overall concordance 89% [95% CI 84-95], sensitivity 91% [82-99], and specificity 89% [81-97], with an AUC of 0·93 [0·89-0·99]) and 1 × FNAB performed the best among the smear protocols, with a concordance of 84% (95% CI 74-93), sensitivity of 63% (43-82), specificity of 97% (92-100), and AUC of 0·91 (0·72-0·97). For HER2, Maui had the highest agreement, with an overall concordance of 93% (95% CI 89-98), sensitivity of 96% (88-100), specificity of 92% (87-98), and AUC of 0·95 (0·98-1·00). For Ki67, Maui had the best performance of smear protocols, with a concordance of 73% (95% CI 64-82), sensitivity of 70% (58-81), specificity of 81% (66-96), and AUC of 0·80 (0·54-0·82). Interpretation: Processing FNAB samples with STRAT4 is feasible in Tanzania, and performance for the oestrogen receptor is robust. Further optimisation of STRAT4 for FNAB has the potential to improve timely access to breast cancer diagnostics in resource-constrained settings. Funding: US National Institutes of Health; UCSF Global Cancer Program, Helen Diller Family Comprehensive Cancer Center; UCSF Department of Pathology; and Cepheid.