Cell‐free DNA integrity correlates with lymph node metastasis and predicts short‐term postoperative recurrence in patients with proximal gastric carcinoma after curative gastrectomy

The predictive value of cell‐free DNA (cfDNA) integrity for lymph node metastasis and short‐term postoperative recurrence in patients with proximal gastric carcinoma (PGC) after curative gastrectomy remains unclear. From August 2019 to March 2021, the cfDNA integrity was assessed in 107 patients with proximal gastric carcinoma who underwent total gastrectomy (TG). The correlation between cfDNA integrity and lymph node metastasis was inspected, and the optimal cutoff value of cfDNA integrity for lymph node metastasis was determined. Meanwhile, the cfDNA integrity was identically detected in 91 proximal gastric cancer patients receiving proximal gastrectomy (PG), as well as 41 heathy individuals. The diagnostic value of cfDNA integrity for short‐term postoperative recurrence was assessed in both cohorts. The cfDNA integrity in healthy control was significantly higher than that in TG (p < .001) and PG (p = .0003) cohorts, but not statistically different between the latter two (p = .540). In the TG cohort, the cfDNA integrity was closely correlated to lymph node metastasis. Using the cutoff value of 0.873, the patients were stratified into the cfDNA integrity high and low groups; low cfDNA integrity was associated with a shorter disease free survival (DFS) both in the TG (p = .011) and PG (p = .041) cohorts. Moreover, cfDNA integrity was assessed to possess the most predictive value for postoperative recurrence by ROC curves in both cohorts. Therefore, the cfNDA integrity may correlate with lymph node metastasis in patients with PGC, and predict short‐term recurrence after curative gastrectomy.

was unsatisfactory; although the reported survival rates varied from 10% to 58.2%, 6-10 the cumulative survival of PGC was significantly lower than cancers originated from other parts of the stomach. 2 The poor prognosis is considered to be the result of local infiltration and lymph node invasion, 11 which might contribute to early recurrence. So far, few methods could offer an accurate diagnose of the disease whether primary or recurrent, even computed tomography (CT) and laparoscopy were assessed to be only partially beneficial. 12,13 Therefore, novel markers are imminently needed for PGC.
The utility of cell-free DNA (cfDNA) in cancer management has been broadly proved. 14 The clinical application of cfDNA is diverse and each with its advantages. While the concentration of cfDNA is easy to acquire and the genetic mutations harbored within cfDNA offer more comprehensive and stable information, the integrity of cfDNA possesses both of the features. It is hypothesized that during the release from apoptotic cells, cfDNA was truncated into 200 bp fragments evenly, but when originated from tumor sites, cfDNA more likely consist of higher proportion of shorter debris since the source is also comprised of necrotic cells. 15 So, by measuring the ratio of longer to shorter DNA fragments, the cancer-related situations might be revealed. 16,17 Recently, cfDNA was reported to be predictive of axillary lymph node metastasis in patients with breast cancer. 18 Hence, we assessed the underlying correlation between cfDNA and local lymphatic metastasis as well as short-term postoperative recurrence in patients with PGC after resections.  19,20 The detailed inclusion criteria were as follows:

| Patients and samples
(1) diagnosis of PGC and confirmed by postoperative histopathology, (2) total gastrectomy (TG) surgery for TG cohort and proximal gastrectomy (PG) surgery for PG cohort, and (3) access to complete clinical data. The patients were followed-up every 1-3 months after discharge, until recurrence, death, or the end of the study. Healthy control was comprised of individuals receiving regular exams.
Blood samples were collected on the morning of admission using 10 mL EDTA-coated tubes (BD) through ulnar vein. The samples were handled within 2 h after drawn.

| cfDNA extraction and integrity calculation
A two-step procedure was adopted to process the samples. A 10-min centrifugation at 3000 rpm was performed to separate the plasma from the hemocytes and the plasma was furtherly centrifugated at 12 000 rpm to remove the cellular fragments. The QIAamp Circulating Nucleic Acid Kit (QIAGEN) was employed to extract cfDNA out of 1 mL plasma. The particular method was documented in a previous study. 21 The purified plasma was stored at À80 C before and after cfDNA extraction.    Figure 1A).

| Predicting value of cfDNA integrity for lymph node metastasis and prognosis in the TG cohort
Then, we observed a close correlation between cfDNA integrity and lymph node metastasis in the TG cohort. Patients with lower cfDNA integrity tended to have more metastatic lymph nodes ( Figure 1B).
Likewise, patients with lymph nodes metastasis were associated with significantly lower cfDNA integrity than patients with non-lymphatic metastasis ( p = .0015, Figure 1C). We furtherly analyzed the diagnostic value of cfDNA integrity for lymph node metastasis in proximal gastric carcinoma undergoing TG, the ROC curve was generated, and the area under the curve (AUC) was calculated. According to the ROC, the Youden index was maximal when the cfDNA integrity was 0.873, which achieved a sensitivity of 43.3% with the specificity being 90.9% ( Figure 1D). Therefore, we divided the TG cohort into two groups using the cutoff value of 0.873. Logistic regression proved cfDNA integrity to be the only risk factor for lymph node metastasis among the preoperative indexes (Table 3). In the meantime, the AUC of cfDNA integrity for lymph node metastasis also remained the greatest of all the preoperative clinical indexes ( Figure 2E) and the disease free survival (DFS) in the high cfDNA integrity group was statistically longer than that in the low cfDNA integrity group (unreached vs. 7 months, p = .011, Figure 1F). Furthermore, cfDNA integrity, together with positive lymph node, was assessed to be the independent risk factors for postoperative recurrence in the TG cohort by Cox regression (Table 4).

| Predictive value of cfDNA integrity for lymph node metastasis and postoperative recurrence in the PG cohort
In the PG cohort, the distribution of cfDNA integrity and number of lymph node metastasis is presented in Figure 2A. Thus, we divided the patients into high and low cfDNA integrity groups by the cutoff point of 0.873. The logistic regression showed a positive correlation between cfDNA integrity and lymph node metastasis (Table 5). Moreover, Cox regression suggested microvascular invasion (MVI) and cfDNA integrity F I G U R E 1 Correlations between cfDNA integrity and lymph node metastasis/postoperative recurrence. (A) Comparison of cfDNA integrity among healthy group and PGC groups: cfDNA integrity in healthy group was significantly higher than the PCG group, while the PCG group presented close cfDNA integrity values; (B) Correlation between number of metastatic lymph nodes and cfDNA integrity, Y axis stands for the number of the metastatic lymph nodes (orange) and the cfDNA integrity counts (blue). X axis shows the patients according to the number of lymph node metastasis; (C) Comparison of cfDNA integrity between patients with and without lymph node metastasis in the TG group: patients with lymph node metastasis had a lower cfDNA integrity than patients without lymph node metastasis; (D) ROC curve of cfDNA integrity for lymph node metastasis; (E) ROC curves of preoperative indexes for diagnosing lymph node metastasis: The AUC of cfDNA was 0.671 and ranked the first among all the indexes; (F) Kaplan-Meier analysis of patients with high and low cfDNA integrity in the TG cohort.
T A B L E 3 Logistic regression for lymph node metastasis in the TG cohort. to be the independent risk factors for postoperative recurrence in the PG cohort (Table 6). Furthermore, the AUC of cfDNA integrity (0.614) also ranked the first among all the preoperative clinical indexes for lymph node metastasis ( Figure 2B), and patients with high cfDNA integrity had a significant longer DFS than patients with lower cfDNA integrity (unreached vs. 9 months, p = .041, Figure 2C).

| DISCUSSION
According to previous research, recurrent tumor within 1 year after treatment has a high chance of originating from the primary sites, 24 and also greatly compromise the treatment outcomes. Unfortunately, there are still very limited ways to monitor early relapse in cancer patients. Recent years, cfDNA has demonstrated great potential in cancer management because it is stably detectable, rich in connotation, and diversely researchable. There are various methods of utilizing cfDNA for clinical oncological research. The most common ones are by detecting concentration, integrity, or sequencing. Although cfDNA concentration is easy to obtain, the result may be greatly biased either by the patients' body state or by the preanalytical procedures. 25 Digging the genetic alterations within cfDNA by next-generation sequencing could be informative, 26 but the high cost limited its application. In contrast, cfDNA integrity is determined by the ratio of long and short fragments, and free from the interference caused by physiological fluctuation. Moreover, cfDNA integrity was proved to be a promising biomarker for early GC, and its perioperative change pattern reflected the therapeutic effect. [27][28][29] In our study, although the patients showed no significant difference in gender, total hospital stay, and surgical complications in the two cohorts, the patients in the TG cohort were associated with older age, higher CEA level, larger tumor size, more advanced clinical stage, and more incidences of MVI and lymph node metastasis compared to the PG cohort. The differences between the two cohorts were likely the results of the strategy of surgical choice in our center. We tended F I G U R E 2 Correlations between cfDNA integrity and lymph node metastasis/postoperative recurrence in the PG cohort. (A) Comparison of cfDNA integrity between patients with and without lymph node metastasis in the PG group: patients with lymph node metastasis had a lower cfDNA integrity than patients without lymph node metastasis; (B) ROC curves of preoperative indexes for diagnosing lymph node metastasis: The AUC of cfDNA was 0.614 and ranked the first among all the indexes; (C). Kaplan-Meier analysis of patients with high and low cfDNA integrity in the PG cohort.