Construction of a model for evaluating the efficacy of neoadjuvant chemotherapy for breast cancer and dynamic monitoring of ctDNA response to neoadjuvant chemotherapy.

e12600 Background: Neoadjuvant chemotherapy (NAC) is a routine treatment of choice for patients with locally advanced breast cancer. The pathological complete response (pCR) to NAC in breast cancer is closely related to a better prognosis. In addition, there have been few studies of the role of ctDNA in the dynamic monitoring of NAC, so we explored the prediction model of NAC to predict pCR and evaluated the role of ctDNA in the dynamic monitoring of NAC. Methods: A total of 269 breast cancer patients receiving NAC were enrolled, and a total of 266 tissue samples were collected. The tissue samples were sequenced using a panel covering 457 cancer-related genes to construct a pCR prediction model after NAC. A total of 267 blood samples were collected from 56 patients. Blood samples were collected at the indicated time points: before NAC (T0), after the first NAC and before cycle two (T1), during intermediate evaluation (T2), and after the end of NAC but before surgery (T3). We constructed a model to predict pCR after NAC by mutated genes and clinical factors, analyzed ctDNA of blood samples according to the mutated genes of the prediction model, and detected the dynamic monitoring role of ctDNA in NAC to predict prognosis. The median follow-up time for survival analysis was 898 days. Results: A total of 192 patients were enrolled to construct the prediction model. There were 51 patients in the additional validation set. We analysed the somatic mutations of 192 samples and constructed a predictive NAC response model including 5 SNV mutations (TP53, SETBP1, PIK3CA, NOTCH4 and MSH2), 4 CNV mutations (FOXP1-gain, EGFR-gain, IL7R-gain, and NFKB1A-gain), and 3 clinical factors (luminal A, Her2+ and Ki67). Analysing the ctDNA of 267 blood samples through a unique panel composed of 9 mutant genes in the prediction model, it was found that ctDNA positivity decreased with the passage of time during NAC, the ctDNA positive rates of ctDNA from T0, T1, T2 to T3 were 46%, 14%, 13% and 10%, respectively. According to survival data, pCR patients did not have disease progression after NAC. Among the non-pCR patients who had disease progression, the probability of non-pCR in patients with ctDNA cleared at T1, T2, and T3 was significantly lower than that in patients with uncleared ctDNA. Interestingly, patients who failed to achieve pCR but were ctDNA negative had a similar risk of metastasis and recurrence as those who achieved pCR. Finally, we found that the prediction model combined with ctDNA could predict pCR after NAC. It had high sensitivity and specificity and the AUC value reached 0.961. Conclusions: This study established a predictive model for predicting pCR after NAC. At the same time, ctDNA dynamic monitoring found that ctDNA status could predict NAC response and metastasis recurrence. Combining the prediction model and ctDNA status could better predict the NAC results.