Hormone receptors, HER2 and Ki67 are considered now as a crucial step in the management and prognosis for breast cancer (BC). mRNA based approach for evaluation of these markers could present a more valuable and accurate method than the current used modality. The Xpert® Breast Cancer STRAT4, is the prototype device for this method. We compared mRNA expression analysis of ER, PR, HER2, and ki67, with central immunohistochemistry (IHC) by recruiting 100 breast cancer patients from Kasr El Ainy Hospitals, Cairo University, Egypt.
In our study, there was a high overall concordance rate between the STRAT4 results compared with IHC results concerning the four biomarkers ( 96% for ER, 94.7% for PR, 93% for HER2 & 97.8% for Ki67) & the Kappa measure of agreement showed perfect agreement (κ = 0.92 for ER, 0.88 for PR, 0.81 for HER2 & 0.91 for Ki67). While the sensitivity (positive percent agreement) was 92.7% for ER, 96.6% for PR, 81.5% for HER2 & 98.8% for Ki67 and the specificity (Negative percent agreement) was 100% for ER, 91.7% for PR, 97.3% for HER2 & 92.9% for Ki67.
Wu et al., 2018  & Fillipits et al., 2021  in their studies to compare the two modalities, show also a high concordance rate of ER & was close to ours (97.8% & 98.9%) respectively, with also nearly the same Kappa agreement (91.8%) in Wu et al., 2018  study, but a slightly lower results were reported in Denkert et al., 2019 , lower concordance rate (89.7%), lower Kappa coefficient (79.38%).
While concerning PR, most of the studies showed slightly lower results related to comparing IHC & STRAT4 as seen in Fillipits et al., 2021 , Janeva et al., 2021  & Wu et al., 2018  with concordance rates (89.9%, 89% & 90.4%) respectively.
For HER2, A close figure to our results was mentioned in Wasserman et al., 2017 , with a concordance rate 91.25%, specificity of 94% & slightly higher sensitivity (87%). A higher concordance rate was seen in Fillipits et al., 2021  & Mugabe et al., 2021  (98.2% & 97.8%) respectively. As noticed in ER & PR, Denkert et al., 2019  in their study showed also a lower concordance rate for HER2 than ours (81.7%).
In most of the literature, results concerning the Ki67 were lower than our study as Fillipits et al., 2021  & Janeva et al., 2021  showed concordance rates (84.4% & 76%) respectively, but the same high sensitivity was reported in Denkert et al., 2019  (97.8%), yet the specificity was very low 30.6%.
Several points could offer an explanation for these differences seen between studies and ours. First, some studies used old blocks archived for long periods as in Fillipits et al., 2021  which used samples from 1990 to 1995 which were incorporated in the ABCSG trial 6. Also in Wu et al., 2018  study, some of the blocks used aged 22 years. This long period can lead to mRNA degradation, but our study was prospective, whereas the patients are chosen then the samples are collected and tested.
Second, the type of biopsy used for testing can offer a further explanation, as some studies used CNB, others used surgical excision specimens and some tested on both types. And this can readily affect the results especially when using surgical excision specimens with extensive DCIS or tissue sections with large amount of adjacent normal breast tissue and other precursor lesions which will affect the results particularly if the tumor is ER/PR negative. This was avoided in our study as we used CNB specimens & further examining all the samples by standard H&E pathology sections, determining, locating & scraping the exact invasive tumor tissue and excluding any normal tissue or non-invasive tumor.
Another possible factor that might account for the higher concordance documented in this study is the amount of tumor tissue in the sample tested, as if the tumor tissue is insufficient, this will render the STRAT4 test indeterminate or negative especially for PR & Ki67 so we tried harder to choose samples that contain enough invasive tumor tissue and excluding the few indeterminate cases from our analysis.
Furthermore, the usage of different clones for IHC as well as the different cut off values used with Ki67 between different studies, different methods of interpretation & subjective variability in IHC interpretation can be a key point for figuring out theses variances.
Regarding the post-neoadjuvant systemic therapy response, 28% reached pCR, 55% showed only partial response, 8% showed stable or progressive disease while 9% was metastatic. Comparing this to Silver et al., 2010  showed different results, only 17.8% showed complete response, 50% showed partial response & 32.2% showed either stable or progressive course. This difference can be explained by the small sample in Silver et al., study (only 28 patients). Also Zhao et al., 2015  & Olfatbakhsh et al., 2018  showed lower rate of complete response among their study groups, (14% & 19%) respectively.
The rates of pathological complete response differed between the molecular subtypes, with 21.7% for luminal A, 15% for luminal B, 47.4% for HER2/neu-enriched, and 44.4% for triple-negative subtype. These were near to that reported by Zhao et al., 2015  in his study.