High concentration of plasma cell free DNA alerting disease recurrence in high risk neuroblastoma children

Background Neuroblastoma is the third-most common cancer in children. The high rate of tumor recurrence accounts for a low survival rate in high risk neuroblastoma. Therefore it is clinically of extreme importance to find an effective biomarker for alerting disease recurrence. Methods Total 116 high risk neuroblastoma patients were recruited in Beijing Children’s Hospital from February, 2015 to December, 2017. All patients had received multiple-disciplinary treatment, then went into maintenance treatment phase after evaluation. Blood samples were collected to quantify plasma cell-free DNA (cfDNA) at time points of the beginning of maintenance treatment, every three months afterwards, and diagnosis of recurrence. Results Results showed that 36 high risk neuroblastoma patients developed recurrence during maintenance treatment. The plasma cfDNA concentration was significantly higher in recurrence than in event-free patients (29.34 ng/ml VS 10.32 ng/ml). The time span of cfDNA level higher than 29 ng/ml was consistently detected ahead of recurrence at mean of 0.55 months. The ROC analysis showed that AUC was 0.825, optimal sensitivity and specificity of 80.6% and 71.3% respectively, at cfDNA level of 12.93 ng/ml. Conclusions We concluded that high level of plasma cfDNA could serve as a promising molecular marker to alert recurrence disease in high risk neuroblastoma children .

abdomen, but might also develop in the neck, thorax, or pelvis. Symptoms and signs included various neoplasms in the abdomen, neck, or thorax, and bone pain. [6,7] Treatment of NB was based on risk stratification, typically including surgery, chemotherapy, radiation, immunotherapy, et cetera. [8][9][10][11] The toughest thing for physicians is tumor recurrence in high risk NB patients, as results in a survival rate less than 50%. [12][13][14] Currently, imagological and cytological examinations could identify recurrent diseases and metastatic tumor sites. [15] But the recurrence could not be identified until tumor cells re-grew significantly. Consequently, patients missed out on an optimal treatment time window. Therefore it is of extreme importance to find out an effective biomarker for alerting recurrence.
Nowadays plasma cell free DNA (cfDNA) has become an active and promising biomarker in cancer research. [16][17][18] Generally, cfDNA is degraded DNA fragments which enter the bloodstream during tumor apoptosis or necrosis. Those DNA fragments are normally cleaned up by macrophages, but some cfDNA is left in blood because of overproduction of tumor cells. This gave a hint that cfDNA could be used to monitor cancer development in clinic. [3,19,20] Few researches had been carried out so far to identify cfDNA as a biomarker for NB recurrence. In this research, we monitored cfDNA levels in high risk NB patients during maintenance treatment to detect tumor burden and alert recurrence. We found that high level of plasma cfDNA was closely correlated with tumor recurrence, and it could potentially be a satisfying molecular marker in clinic.  [21] and a German report. [22] Diagnostics and Evaluation Patients of CR, PR and SD could step into maintenance treatment.

Patients
Evaluation during maintenance treatment included serum tumor markers detection, microscopic examinations of bone marrow and imaging examination every three months, and 131 I-MIBG every six months.

Treatment
According to the BCH-NB-2007-HR protocol, initially diagnosed high risk NB patients received multiple-disciplinary treatment including induction chemotherapy, surgery, consolidation therapy, and radiotherapy. Some patients received autologous stem cell transplantation. Regular regimens included chemotherapy with high dose cyclophosphamide, adriamycin and vincristine (CAV), chemotherapy with high dose cisplatinum and VP16 (CVP), surgery after 4-5 cycles of chemotherapy, and then harvest peripheral blood stem cell for possible autologous hematopoietic stem-cell rescue.

Sample collecting
Blood samples were collected to quantify cell-free DNA (cfDNA) at time points of the beginning of maintenance treatment, every three months afterwards, and diagnosis of recurrence. Venous blood samples were collected in EDTA-coated tubes and plasma was separated by centrifugation at 1600g for 10 minutes. Supernatant was transferred into a new tube and centrifuged at 16000g for 10 minutes. Purified plasma was carefully removed from the tube and stored at -80°C for DNA extraction. Cycling conditions were 1 minute at 95°C and 35 cycles of 95°C for 8 seconds and 60°C for 15 seconds. The qPCR reaction was performed in triplicate and the mean of the triplicates were used for further analysis. Each plate contained as a plasma DNA sample, a water template (negative control) and 7 serially diluted standard DNA solutions. The cfDNA level of each sample was detected based on standard curve by using the 2 -△△Ct method.

Statistics analysis
Statistical analysis was performed in R statistical environment (version 3.4.0), including Mann-Whitney U test, Chi-Square test, and ROC (receiver operating characteristic, Bioconductor ROC package). A p-value lower than 0.05 was considered statistically significant.

Demographic and clinical characteristics
Eligible 116 high risk NB pediatric patients at maintenance treatment were enrolled ( Table   1). All 7 high risk patients younger than 18 months were detected amplification of MYCN gene. Ninety-four percent of patients were older than 18 months, one of those with amplification of MYCN gene was classified as stage III by INSS and others as stage IV.
There was no significant difference of proportion between female and male. At initial diagnosis, 80.2% of the patients had primary tumor site in abdomen, 17.2% in thorax, and 2.6% in others. Around 75.9% of the patients had NSE level blow 370 ng/ml. while 75.9% of the patients were detected LDH concentration between 500 IU/L and 1500 IU/L, 12.1% lower than 500 IU/L, and others higher than 1500 IU/L. Remarkably, 50.9% of the patients had less than 3 organs with metastasis, 33.6% three organs, and 15.5% more than three organs. The most frequent metastatic sites were bone, bone marrow and distant lymph node, accounting for 72.4%, 62.1% and 65.5% respectively.

Recurrence during maintenance treatment
All 116 patients stepped into maintenance treatment after evaluation, when their multipledisciplinary treatments finished, with results that two independent microscopic examinations confirmed no NB cells in bone marrow, independent radiological experts confirmed no progression via radiography, LDH and NSE were down-regulated, and 131 I-MIBG scan negative.
During maintenance treatment, recurrence was diagnosed by positive microscopic examinations of bone marrow and tumor finding in situ or metastasis via radiography. A total of 36 patients developed recurrence during maintenance treatment ( Table 2). Five cases occurred in the first-three months with 13-cis-retinoic acid, and 26 cases in the next 6~9 months. Other 5 cases happened after 13-cis-retinoic acid was stopped.

Analysis of cfDNA level in recurrence and event-free patients
As presented in Figure 1, plasma cfDNA concentration of recurrence patients at time point immediately ahead of diagnosis of recurrence was significantly higher than the average level of event-free patients during a whole maintenance treatment (median, 29.34 ng/ml VS 10.32 ng/ml, p<0.001). Whereas the baseline comparison showed that the average cfDNA level of recurrence patients pre-peak concentration (pre-recurrence) was not significantly different from the average level of events free patients during a whole maintenance treatment (median, 9.75 ng/ml VS 10.32 ng/ml, p>0.05, Figure S1). Nineteen recurrence patients had cfDNA peak level higher than the median of 29 ng/ml (Table 3). Time span from time point of cfDNA peak level to diagnosis of recurrence ranged from 0 to 3 months, at mean of 0.55 months. This suggested that high level of plasma cfDNA could potentially be a timely molecular marker to alert recurrence.
The ROC analysis showed that AUC (area under ROC curve) was 0.825, optimal sensitivity and specificity of 80.6% and 71.3% respectively, at cfDNA level of 12.93 ng/ml (Figure 2).
It suggested that high level of plasma cfDNA was significantly correlated with recurrence disease. The consistence between high level of cfDNA and NB recurrence elucidated the correlation between heavier tumor burden and high level of cfDNA.

Recurrence among subgroups of NB patients during maintenance treatment
All patients were classified into three subgroups (i.e., CR, PR, SD) based on the evaluation of therapeutic response to multiple-disciplinary treatment before they stepped into maintenance treatment. As shown in Table 4, 12% of CR patients developed recurrence during maintenance treatment, 45.5% in PR and 45.5% in SD subgroups. The recurrence rate of CR patients was significantly lower than in PR or SD. Meanwhile, the time span from beginning of maintenance to diagnosed recurrence in CR subgroup was significantly longer than it in PR or SD (mean, 17.52 months VS 10.97 and 6.64 months respectively).

Discussion
Neuroblastoma is one of the most common cancer in children, with 7% increment of incidence during 10 years. [22,24] Although clinical therapy including radiological imaging, cytological, biochemical and molecular techniques, et cetera, progressed immensely, long-term survival rate of high risk NB patients was still below 50%.[8, 25,26] Many factors contributed to this disappointed result, and the most notable one was chemo-resistant minimal residual disease which caused recurrence in more than half of high risk NB patients. [2,14,24,27] Therefore, it is on top priority for physicians to identify minimal residual disease accurately, to evaluate therapeutic response reliably, and to establish alerting system effectively.
Nowadays many studies focused on finding promising biomarkers to evaluate prognosis or response to treatments in cancer therapy. Plasma cfDNA as a potential biomarker in clinic was investigated profoundly recent years, especially in malignant metastatic cancers. Our previous study demonstrated that plasma cfDNA level was highly correlated with tumor burden in NB children, and could potentially serve as a more effective biomarker compared with LDH widely used in clinic. 21 Furthermore, plasma cfDNA concentration was significantly down-regulated in partial remission patients than in stable disease, and dynamically associated with tumor burden in response to chemotherapy.
[34] However, whether cfDNA could serve as an effective molecular marker for recurrence had yet to clarify.

Conclusions
In our current work, we found that plasma cfDNA level of recurrence patients at time point immediately ahead of diagnosis of recurrence was significantly higher than the average level of event-free patients during a whole maintenance treatment (median, 29.34 ng/ml VS 10.32 ng/ml, p<0.001, Figure 1). While the average cfDNA level of recurrence patients before time point of peak level had no significant difference from the average level of event-free patients during a whole maintenance treatment (median, 9.75 ng/ml VS 10.32 ng/ml, p>0.05, Figure S1). Then we investigated whether high level of cfDNA had the potential to serve as an effective molecular marker to alert recurrence in advance. As shown in Table 3, time span from time point of cfDNA peak level (more than 29 ng/ml) to diagnosis of recurrence ranged from 0 to 3 months, at mean of 0.55 months. This indicated that plasma cfDNA could serve as a promising molecular marker to alert recurrence disease in NB patients. The performance of ROC analysis showed that AUC was 0.825, optimal sensitivity and specificity of 80.6% and 71.3% respectively, at cfDNA level of 12.93 ng/ml ( Figure 2). It further suggested that high level of cfDNA was significantly correlated with heavier NB tumor burden, such as in recurrent disease.
In conclusion, high concentration of plasma cfDNA could be one promising timely molecular marker to alert recurrence disease for high risk NB patients during maintenance treatment, at least an effective assistance in clinic. The level of cfDNA between NB patients in recurrent and event-free cases. The p value was <0.001, analyzed by Mann-Whiney U Test).

Figure 2
The performance of cfDNA to evaluate recurrent disease in NB patients.

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.  Figure S1.tif Table 3.xls Table 4.xls