*For medical professionals for reading reference
Studies show that AI is evaluating The scoring is highly consistent with pathologists, and can improve the ability to identify patients in HER2 ultra -low expression, providing a new perspective for the diagnosis and treatment of breast cancer.
Breast cancer is a kind of malignant , which is the first place in female malignant tumors [1]It has a serious threat to womens physical and mental health.The final test results of HER2 immunohistochemistry are affected by multiple quality control links before, during and tested. Therefore, how to improve quality management technology in HER2 IHC detection has become a highly concerned research hotspot in recent years.At the U.S. and Canadian Society (USCAP) conference held in 2024, A study using artificial intelligence (AI) assisted assessment of breast cancer hollow needle puncture biopsy (CNB) specimen HER2 ultra -low expression and HER2 low expression immune groupThe results show that the surgical resection biopsy (SEB) and the HER2 low expression or HER2 ultra -low expression state in CNB may change, and the AI auxiliary and artificial scores are highly consistent [2] . This study reminds the necessity of re -detecting HER2 in the resection specimen, providing evidence based on HER2 standardized detection and diagnosis.The important content is sorted as follows to readers.
Research background
New antibody drugs (ADC) have expressed HER2 low -expression breast cancer patientsThe clinical treatment strategy has a significant impact.At the same time, pathologists will face new challenges, such as accurately distinguish HER2 immunohistochemistry (IHC) scores of 0 and 1+.Generally, HER2s status is determined by CNB.Therefore, this study aims to use AI technology to evaluate whether the HER2 low expression and ultra -low expression status in CNB can reflect the HER2 status of the entire tumor.
Research and Design
This study was incorporated with retrospective analysis from January 2021 to December 2021 AcceptedThe corresponding CNB specimen and SEB samples of patients with infiltrated breast cancer patients.Study the use of automated dyeing equipment and specific antibodies against HER2 for dyeing, and then score through the AI tool.The training of AI tools is based on 265 slices of 6012 patch test data sets.All HER2 IHC slices are checked and scored by two pathologists who have been trained and scored 0 (including zero and ultra -low), 1+, and 2+.
Figure 1. Research design
Research Results
AI is consistent with artificial assessment
AI and manual re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -re -reinity manual manual manual manual.The overall consistency between the scoring results is 91.49%, showing that AI has the ability to identify with well -trained pathologists in HER2 IHC 0 and 1+ scores.
Figure 2. Her2 AI algorithm to IHC 0/1+recognize the consistency rate high
AI can improve the recognition of patients with HER2 ultra -low expression
CNB re -evaluation results show that IHC 0 and IHC 0 andThe consistency rate of IHC is 95.26%between historical and re -scoring results (kappa = 0.678, P & LT; 0.001).
AI score results show that the consistency rate between IHC 0 and IHC low in AI and artificial re -scoring results is 90.51%(kappa = 0.465, P & LT; 0.001).It is worth mentioning that 32 patients were scored in HER2 ultra -low expression by the AI scoring, of which only 4 of them were rated as ultra -low expression by artificial scores, showing that the AI tools could improve the recognition of patients with HER2 ultra -low expression.
Figure 3. AI can increase the identification of patients with HER2 ultra-low expression
The consistency of artificial and AI CNB and SEB scores
The results of artificial scores show that the overall inconsistency rate of CNB and SEB samples is 22.13%.In the SEB sample, patients with 28.79%(19/66) have undergone the transformation of HER2 negative/ultra -low expression to HER2 low expression state, and patients at 21.14%(93/440) have gone through HER2 low expression to HER2 negative/ultra low low.Transformation of expressions.
AI auxiliary scores show that the overall inconsistency rate of CNB and SEB samples is 24.11%.In the SEB sample, patients with 21.88%(7/32) have undergone the conversion of HER2 ultra -low expression to HER2 low expression state, while 115 patients with (24.26%) have experienced HER2 low expression to HER2 negative/ultra -low expression stateTransformation.Among the 474 patients with HER2 1+ into HER2 -negative/ultra -low expression state, the percentage of influential tumor cells with weak and incomplete membrane dyeing on CNB glass was lower than HER2 low expression state (P & LT; 0.001).
Figure 4. Similar consistency of artificial and AI CNB and CEB.Research conclusions
This study proves the potential of AI in the evaluation of the HER2 IHC score of the auxiliary pathologists.AI tools can not only improve the consistency and accuracy of the score, but also assist pathologists to identify HER2 ultra -low expression patients, which is essential for the formulation of personalized therapy plans.This study provides a new perspective for the diagnosis and treatment of breast cancer.The AI auxiliary assessment tool is expected to become a powerful assistant for pathologists, helping them to evaluate HER2 status more accurately, thereby providing patients with more personalized treatment suggestions.With the continuous progress of AI technology, its application in the field of pathology will become more and more extensive.Future research may further explore the application of AI in the evaluation of other cancer markers, as well as how to integrate AI tools into the clinical work process to improve diagnostic efficiency and precise treatment effects of patients.
References:
[1] BRAY F, LAVERSANE M, SUNG H, et al. Globalcancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024 Apr 4.
[2]Li M, Cai M, Lv H, ET Al. Artificial Intelligence Assisted Assessment of Her2-Ultralow and Her2-LOW ImmunoHistochemical Scoring In Breast Cancer Core Biopsy Specimens. 202 4 uscap: abstract #178.
Wonderful informationWaiting for you
Approval Number: CN-146625 Valid until: 2025-01-31
Provided by Like for medical professionals for reference only
