Utility of urinary ctDNA to monitoring minimal residual disease in early breast cancer patients

2.1 Study design

Serial urine samples were recovered from early breast cancer patients and healthy volunteers at different time intervals. Patients were recruited at the Xiangyang Central Hospital and treated with standard therapy. Healthy volunteers were recruited from individuals from health screening programs and were deem disease free from mammograms. All procedures were approved by the institutional review board and strictly follows the Declaration of Helsinki. Written informed consent was obtained from all trial participants. A total of 300 patients and 50 healthy volunteers were recruited. Healthy controls were recruited at healthcare screening programs with no abnormalities detected via a standard mammogram. For early breast cancer patients, majority were walk-in patients with complains of lumps or pain in their breast region. Mammogram showed positive results and followed up by CT scans as part of routine diagnostic workup. As per the adjudication of the attending physician, no metastatic disease was detected, and these patients were then recruited. Patients with metastatic disease were excluded. Breast cancer patients were treated with neoadjuvant chemotherapy. After completion of surgery, patients were provided with post-operative radiotherapy. Urinary ctDNA was assessed prior to treatment and compared with tumor biopsy results. Post-surgery samples were also taken. Thereafter, measurement time points were taken at monthly intervals for 6 months. Patient follow up via clinic visits or phone calls were conducted thereafter for a period of 3 years at monthly intervals for correlation of relapse. Throughout the study, the attending physician and the study subject were not made known of the results and standard of care for patients were not altered in any way for the best interest of the patients.

2.2 Clinical sample preparation

Core biopsy were taken as per routine practice for histopathological analysis. For urine samples, the first morning midstream urine were collected and processed within 3 hours of collection. 50 mL of samples were taken in a sterile receptacle. For urine processing, urine samples were transferred to a 50 mL centrifuge tube and spun down at 4,000 g for 10 minutes at 4 ∘ C. The supernatant was carefully transferred to a new 50 mL centrifuge tube and a repeat centrifugation at the same setting was performed. This ensured any cellular components or debris were removed prior to nucleic acid purification.

2.3 Purification of urinary DNA and ddPCR

Molecular profiles of PIK3CA and TP53 for each cancer patient was established from tissue biopsies. For urinary ctDNA comparison, we utilized ddPCR to sensitive readout. For nucleic acid purification from urine specimens, the Qiagen QIAamp cell free Circulating Nucleic Acid kit (Qiagen Inc, USA) was used. The kit utilizes a series of centrifugation steps and spin columns and processes strictly followed manufacturer’s instructions. The elute containing urinary nucleic acid was collected in 20 μ L TRIS-EDTA buffer. For molecular profiling, primers for PIK3CA and TP53 were procured from BioRad PrimePCR primers (BioRad, USA). Nucleic acid template from each specimen was added to the mastermix and droplet generation performed on the QX100 droplet generator (BioRad, USA). The reaction was partitioned into approximately 14,000 droplets per sample and amplified on a thermocycler at the following conditions. 95 ∘ C for 10 min, followed by 40 cycles of 95 ∘ C for 15 s and 60 ∘ C for 60 s, 10-min at 98 ∘ C. The reaction plates were read on the droplet reader (QX100 droplet reader, BioRad USA). At least two negative controls were incorporated in each plate with blank runs. Amount of quantifiable ctDNA was collated from the QuantaSoft software (BioRad, USA). For total cell free circulating DNA, this is quantified using Nanodrop 2000 (Thermo Fisher Scientific, USA).

2.4 Statistical analysis

The primary goal of the study is to address early breast cancer monitoring especially in MRD and chart its variations during treatment. A direct comparison using Student t test was made between the categorical variables of healthy controls and cancer patients for detecting urinary DNA. This demonstrates the clinical significance in circulating DNA released by the body. One way analysis of variance (ANOVA) was performed to compared the results of quantifiable DNA at different time points. Concordance measure, sensitivity, specificity, positive predictive value and negative predictive value were tabulated for the assay. Multivariate COX regression model and the disease-free relapse curve were computed that will address any confounding factors. The hazard ratio (HR) was determined to assess various parameters linked to risk prediction for early breast cancer.

OPEN IN VIEWER

3.1 Urinary DNA profile at baseline

In a prospective analysis of 350 participants, we investigate the potential utility of urinary ctDNA in early breast cancer. Details of trial participants are shown in Table 1. Among the study group, mean age for patients and healthy volunteers were 53 and 50 years old respectively. Breast cancer patients were profiled for estrogen receptor (ER), progesterone receptor (PR) and HER2 status. The proportion of patient harbouring these key genetic changes (ER/PR+; HER2+; Triple negative) were 44.0%, 32.3% and 23.7% respectively. Additional molecular profiling for PIK3CA and TP53 mutations were conducted on clinical specimens for this current evaluation on urinary ctDNA. Of the 300 clinical samples at baseline, 38% percent of the samples had either one or both mutations. Figure 1A highlights the comparison with tumor biopsy results. Agreement with tumor tissues results was 97.3%. The high degree of concordance with clinical specimens demonstrated the excellent accuracy of the assay for early breast cancer patient profiling. All 50 healthy volunteers showed wildtype characteristics for PIK3CA and TP53. Comparisons were made in different study groups on the recovered circulating DNA quantity. Figure 1B shows the results that demonstrated an elevated level within early breast cancer patients. The quantity obtained from patients were 19.2 folds higher than healthy volunteers. Among the patient group, it was interesting to note that patients that harboured PIK3CA and TP53 mutations from urinary ctDNA had higher levels (Fig. 1C) than wildtype patients. A Student shows statistical significance between these two groups with p value 0.001. We postulate that this may be disease related given breast cancer patients with PIK3CA may be related to poorer prognosis [14] and performed longitudinal comparisons.

3.2 Serial Measurements and tracking urinary circulating DNA

Longitudinal measurements allowed comparison of recovered circulating DNA at different time points post-surgical procedures. We concurrently performed the analysis on healthy controls to ascertain the inherent urinary circulating DNA stability. Molecular profiling was also performed as a means to understand MRD within early breast cancer patients after potential curative surgical procedures. Figure 2A shows the measurements of healthy controls that indicated a fairly uniformly mean extracted urinary DNA over a 6-month period. There were three dropouts for this control group and their results were not included in this analysis. A one-way analysis of variance showed no significant differences in measured background quantity of urinary DNA. We do not also see any significant variability within each time point with the highest and lowest CV at 32% and 30% respectively. For patient measurements highlighted in Fig. 2B, a pairwise ANOVA showed significant changes in the mean recovered urinary DNA. Of note, a sharp decline in mean urinary DNA was seen immediately post-surgery. We compared the mean values to healthy controls at each time point and patients’ urinary DNA levels were statistical significant to be higher using a Student t test. Molecular profiling of the urine samples, showed a decrease in detected mutation rates and comparison with the measurements prior to treatment was a reduction of 15% (Fig. 2C). Interestingly, patients with detectable mutations had significantly higher mean level of urinary circulating DNA as shown in Fig. 2D ( p value 0.001) compared with other patients. We observed a further increase in mutation detection among these patients at the final time point as shown in Fig. 2C. This increased from 57% post-surgery to 83% at 6-month. A further increase in urinary circulating DNA was registered and a Student t test confirmed statistical significance ( p value 0.01) over the measurement taken after post-surgery (Fig. 2D). Another important observation was obtained from the mutation profiling of the remaining patients and healthy volunteers. 100% of them remained as wildtype over the entire monitoring period.

OPEN IN VIEWER

OPEN IN VIEWER

The clear differences in various patient groups highlighted the potential to further explore the parameter to address critical challenges in early breast cancer patients. Follow up contact tracings with patients and collecting data of disease relapses allowed the correlations with urinary circulating DNA to be established. We compared data retrieved at baseline and dividing the patient cohort equally based on the level of urinary circulating DNA. Figure 3A shows no significant differences in disease free survival and relapsed cases in both patient groups that were separated by the level of high and low urinary DNA. The hazard ratio (HR) determined for this comparison was 1.03 (95% CI 0.79 to 1.34). For patients with detectable mutations and those of wildtype TP53 and PIK3CA, we observed that the relapsed rates were higher in the former (Fig. 3B). The HR established in the comparison was 1.98 (95% CI 1.48 to 2.63). We repeated the analysis for post-surgery results and observed stronger separations with urinary circulating DNA recovered at this measurement time point (Fig. 3C). The HR as determined was 1.29 (95% CI 0.99 to 1.68) for patient’s disease-free survival cases. Significant statistical difference was consistently observed with patients grouped by their mutational status (Fig. 3D). HR was 1.79 (95% CI 1.26 to 2.53) and slightly lower to the measurements prior to treatment. Given significant changes was also observed at the end of monitoring, we analysed the disease-free survival rates at 6-month. The results indicated that urinary circulating DNA recovered from patients were significantly correlated. The patient group with higher recovered urinary DNA had worse outcomes and HR was 1.65 (95% CI 1.26 to 2.16) as shown in Fig. 3E. For patients with positive mutations detected from urine samples, we consistently observed clear statistical significance ( p value 0.01) compared with wildtype patients as shown in Fig. 3F. HR was 1.87 and 95% CI was 1.38 to 2.54.