CDR1-AS expression and baseline characteristics
The relative CDR1-AS expression in pan-cancer tissues was analyzed based on the MiOncoCirc database [16] and was found abundant in BRCA group (breast invasive carcinoma) (Fig. 1A). The relative expression of CDR1-AS in BC was higher than other common cancers, such as lung cancer and hepatocellular carcinoma, which were leading fatal cancer types in China. The expression of CDR1-AS was explored in the GEPIA database [17] and was found to be significantly downregulated in BC tissues compared to that in normal breast tissues (p < 0.01) (Fig. 1B). A total of 106 patients were included in this study and divided into two groups based on the median expression of CDR1-AS. Thus, both the high and low CDR1-AS expression groups included 53 patients. No significant differences were found in clinico-histopathological characteristics, including age, body mass index (BMI), ER status, PR status, and other factors, between the high and low CDR1-AS expression group (Table 1). Thirty-seven patients achieved pCR, and the total pCR rate was 34.91%. Thirteen events occurred in this cohort: one patient died, ten relapsed or progressed, and two had secondary primary cancer.
CDR1-AS expression in cancer tissues. A Relative CDR1-AS expression in pan-cancer tissues based on the MiOncoCirc database. B Relative CDR1-AS expression in BC and normal tissues based on the GEPIA database. *p < 0.01 (student’s t-test). ACC adrenocortical carcinoma, BLCA bladder urothelial carcinoma, BRCA breast invasive carcinoma, CHOL cholangiocarcinoma, ESCA esophageal carcinoma, GBM Glioblastoma multiforme, HCC hepatocellular carcinoma, HNSC head and neck squamous cell carcinoma, KDNY kidney cancer, LUNG lung cancer, MBL medulloblastoma, NRBL neuroblastoma, OV ovarian serous cystadenocarcinoma, PAAD pancreatic adenocarcinoma, PRAD prostate adenocarcinoma, SARC sarcoma, SECR secretory cancer, and SKCM skin cutaneous melanoma
CDR1-AS expression and pCR outcomes
The median CDR1-AS expression was 0.099 (range, 0.017 to 1.149) and 0.091 (range, 0.002 to 0.614) in the pCR and non-pCR group, respectively (Fig. 2A). Patients with low CDR1-AS expression achieved a higher pCR rate (41.51%) than those with high CDR1-AS expression (28.30%; Fig. 2B), although the difference was not significant (odds ratio [OR] = 0.556; 95% confidence interval [CI] 0.247–1.250, p = 0.156; Table 2). In univariate logistic regression analysis, negative ER status (OR = 0.182; 95% CI 0.077–0.434, p < 0.001) and positive HER2 status (OR = 3.286; 95% CI 1.430–7.549, p < 0.001) favored pCR. Moreover, negative PR status (OR = 0.480; 95% CI 0.209–1.101, p = 0.083) and low BMI status (OR = 0.462; 95% CI 0.197–1.078, p = 0.074) tended to favor pCR. Age (OR = 1.191; 95% CI 0.525–2.700, p = 0.675), clinical tumor stage (OR = 0.486; 95% CI 0.193–1.223, p = 0.125) and ki67 status (OR = 1.506; 95% CI 0.660–3.437, p = 0.331) were not significantly associated with pCR (Table 2).
Features of patients with pathological complete response (pCR) and non-pCR. A The relative CDR1-AS expression in the pCR and non-pCR group. B Clinicopathological features of pCR and no-pCR patients. Two-category data (pCR, yes vs. no; CDR1-AS expression, high vs. low; age, ≥ 50 vs. < 50 years, clinical T-stage, T4 vs. T2–3; clinical N stage, N1–3 vs. N0; ER positivity vs. negativity; PR positivity vs. negativity; HER2 positivity vs. negativity; Ki67 > 30% vs. ≤ 30%; and BMI ≥ 24 vs. < 24 kg/m2) are shown in dark and light blue, respectively. pCR pathological complete response, ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth factor receptor 2, and BMI body mass index
After adjusting for age, clinical tumor stage, ER status, PR status, HER2 status, Ki67 index, and BMI status, multiple logistic regression analysis revealed that low CDR1-AS expression was significantly associated with pCR (OR = 0.244; 95% CI 0.081–0.732, p = 0.012). Meanwhile, patients with lower clinical tumor stage (OR = 0.123; 95% CI 0.029–0.518, p = 0.004), negative ER status (OR = 0.101; 95% CI 0.027–0.375, p = 0.001), and positive HER2 status (OR = 6.668; 95% CI 2.085–21.328, p = 0.001) could achieve pCR more easily (Table 2).
Building and assessment of the multivariate model for pCR prediction
According to prior multiple logistic regression analysis, four predictive features, including clinical T stage, ER status, HER2 status, and CDR1-AS, were selected to build a multivariate predictive model. A nomogram was created for predicting pCR (Fig. 3A). The calibration curves showed a high consistency between the prediction of the nomogram and the actual observed pCR outcomes in our cohort (Fig. 3B). ROC curves and DCA were used to compare the accuracy of different predictive models with or without CDR1-AS. The area under the curve (AUC) was 0.813 (95% CI 0.727–0.898), achieved by adding CDR1-AS to clinicopathological features, which is better than 0.789 (95% CI 0.700–0.877) for clinicopathological characteristics alone (Fig. 3C). Moreover, DCA consistently showed more benefits with the model combining CDR1-AS with clinicopathological variables (Fig. 3D).
Nomogram of the multivariate model for pCR prediction A The nomogram was built using independent predictive factors for pCR. B Calibration curve of nomogram. C Receiver operating characteristic curves of the predictive models with and without CDR1-AS expression (area under the curve, 0.813 vs. 0.789). D Decision curve analysis of the net benefit versus threshold probability. ER estrogen status, HER2 human epidermal growth factor receptor 2, and pCR pathological complete response
Subgroup analysis of pCR rates
Subgroup analysis suggested that pCR outcomes were significantly associated with CDR1-AS in patients aged ≥ 50 years (OR = 0.096; 95% CI 0.013–0.709; p = 0.022) and those with a BMI less than 24 (OR = 0.172; 95% CI 0.038–0.780; p = 0.022), as well as premenopausal (OR = 0.163; 95% CI; 0.027–0.988; p = 0.048), postmenopausal (OR = 0.090; 95% CI 0.010–0.787; p = 0.030), T2-3 (OR = 0.244; 95% CI 0.068–0.880; p = 0.031), stage N1–3 (OR = 0.186; 95% CI 0.055–0.631; p = 0.007), ER-negative (OR = 0.131; 95% CI 0.021–0.805, p = 0.028), PR-negative (OR = 0.072; 95% CI 0.009–0.584; p = 0.014), HER2-positive (OR = 0.156; 95% CI 0.026–0.945; p = 0.043), and ki67 > 30% tumors (OR = 0.133; 95% CI 0.024–0.732, p = 0.020; Fig. 4). No interaction was detected between the clinicopathological variables and CDR1-AS for pCR (Fig. 4).
Subgroup analysis for pCR according to CDR1-AS expression levels pCR pathological complete response, OR odds ratio, CI confidence interval, ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth factor receptor 2, and BMI body mass index
CDR1-AS expression and DFS
The median follow-up time for all patients was 30.02 months. Kaplan–Meier curves and log-rank tests were performed to determine DFS according to CDR1-AS expression level. Compared to the CDR1-AS low-expression group, the high-expression group showed significantly better DFS (N = 106; log-rank p = 0.022; Fig. 5A).
Kaplan–Meier plot estimates of survival outcomes according to CDR1-AS expression levels A DFS was estimated using the Kaplan–Meier plot. B RFS estimated using the Kaplan–Meier plot. C DDFS estimated using the Kaplan–Meier plot. DFS disease-free survival, RFS relapse-free survival, DDFS distant disease-free survival, HR hazard ratio, and CI confidence interval
In the univariate analysis, patients with high expression of CDR1-AS had a substantially better DFS than those with a low expression of CDR1-AS (hazard ratio [HR] = 0.202; 95% CI 0.044–0.924; p = 0.039). Simultaneously, multivariate analysis showed that CDR1-AS expression was an independent prognostic factor for DFS (adjusted HR = 0.177; 95% CI 0.034–0.928, p = 0.041). Moreover, T4 clinical tumor stage (adjusted HR = 5.445; 95% CI 1.294–22.907; p = 0.021) and high ki67 index (adjusted HR = 7.576; 95% CI 1.436–39.973; p = 0.017) were significantly associated with worse DFS (Table 3).
CDR1-AS expression and RFS
The CDR1-AS high-expression group showed significantly better RFS than the low-expression group (N = 106; log-rank p = 0.012; Fig. 5B). In the univariate analysis, patients with high CDR1-AS expression had substantially better RFS than those with low CDR1-AS expression (HR = 0.112; 95% CI 0.014–0.887; p = 0.038). Multivariate analysis showed that CDR1-AS expression was an independent prognostic factor for RFS (adjusted HR = 0.061; 95% CI 0.006–0.643; p = 0.020). Moreover, T4 clinical tumor stage (adjusted HR = 11.078; 95% CI 2.074–59.164; p = 0.005) and high ki67 index (adjusted HR = 9.880; 95% CI 1.666–58.574; p = 0.012) were significantly associated with worse RFS (Table 4).
CDR1-AS expression and DDFS
The CDR1-AS high-expression group was prone to have a better DDFS than the low-expression group (N = 106; log-rank p = 0.050; Fig. 5C). In univariate analysis, patients with high expression of CDR1-AS tended to have a better DDFS than patients with low expression of CDR1-AS (HR = 0.158; 95% CI 0.019–1.317; p = 0.088). Multivariate analysis revealed that CDR1-AS expression was an independent prognostic factor for DDFS (adjusted hazard ratio [HR] = 0.061; 95% confidence interval [CI] 0.006–0.972, p = 0.047). Furthermore, T4 clinical tumor stage (adjusted HR = 24.665; 95% CI 2.601–233.992; p = 0.005) and high ki67 index (adjusted HR = 19.134; 95% CI 1.776–206.098; p = 0.015) were significantly associated with worse DDFS (Table 5).
Add Comment