The effect of the visceral fat area on the predictive accuracy of C‐reactive protein for infectious complications after laparoscopy‐assisted gastrectomy

Abstract Aim To investigate the influence of visceral fat area on postoperative C‐reactive protein levels and whether it affects its ability to diagnose infectious complications after laparoscopy‐assisted gastrectomy. Methods A total of 435 consecutive patients who underwent laparoscopy‐assisted resection for gastric cancer from 2008 to 2017 were reviewed and divided into four groups according to visceral fat area quartiles. We evaluated the relationship between C‐reactive protein and visceral fat area and whether visceral fat area affects the sensitivity and specificity of C‐reactive protein in diagnosing postoperative infectious complications. Results Postoperative C‐reactive protein levels increased with increasing visceral fat areas at every postoperative assessment. Multiple linear regression revealed that levels on postoperative day 3 significantly positively correlated with visceral fat area. Postoperative day 3 levels also showed moderate accuracy for diagnosing infectious complications (area under the curve, 0.78; sensitivity, 0.86; specificity, 0.65), with an optimal cut‐off of 11.8 mg/dL. The sensitivity for predicting infectious complications was low in the 1st visceral fat area quartile group but high in the 2nd, 3rd, and 4th groups (0.43 vs 1.0 vs 1.0 vs 0.94, respectively). By contrast, the specificity was high in the 1st and 2nd group but low in the 3rd and 4th (0.84 vs 0.70 vs 0.54 vs 0.48, respectively). Conclusion Visceral fat area positively correlated with postoperative C‐reactive protein levels and this affected its accuracy in diagnosing infectious complications. A uniform C‐reactive protein cut‐off may not provide accurate predictions in patients with more extreme visceral fat areas.


| INTRODUC TI ON
Gastric cancer is the fourth most common malignancy and the second leading cause of cancer-related death worldwide. 1 Although surgery is the only curative treatment for gastric cancer, perioperative complications occur in 12.6%-23.6% of cases. 2,3 Infectious complications occur in approximately 10% of patients after gastrectomy. 4,5 Recent studies have demonstrated that postoperative infectious complications (PICs) are associated with inferior long-term survival in patients with various types of cancer, including gastric cancer. [5][6][7] Artinyan et al 7 reported that severe PICs are more closely associated with poor prognosis than less severe infectious complications.
Therefore, it is important to detect PICs at an early stage before they progress, because timely diagnosis is key to improving long-term survival after gastrectomy.
There are several approaches to predicting postoperative complications, [8][9][10] including C-reactive protein (CRP) levels after surgery. 11, 12 Shishido et al 13 reported that CRP on postoperative day (POD) 3 predicted infectious complications following gastric cancer resection. By contrast, a meta-analysis found that CRP could not predict PICs after gastroesophageal cancer surgery. 14  influences the predictive ability of CRP. Therefore, this retrospective analysis was conducted to determine the influence of VFA on postoperative CRP levels and whether it affects CRP's ability to accurately diagnose PICs after laparoscopy-assisted gastrectomy (LAG).

| Patients
The retrospective study included 435 consecutive patients who were pathologically diagnosed with primary gastric cancer and underwent LAG at Yamaguchi University Medical Hospital (Yamaguchi, Japan) between January 2008 and December 2017. Patients with other concomitant malignancies and those without preoperative computed tomography were excluded.
The patients were identified using our database and their detailed information was obtained from original medical records.
This study was approved by the institutional review board of the Yamaguchi University Hospital (H28-182).

| Measurement of body composition parameters
We analyzed the subcutaneous fat area (SFA), VFA, and skeletal muscle area (SMA) on the preoperative multidetector computed tomography (MDCT) images using AZE Virtual Place Raijin software (Aze Ltd). VFA and SFA were measured using axial slices at the level of the umbilicus on preoperative multidetector computed tomography. Fat area was manually measured in the region with Hounsfield units (HU) within the range of −200 to −50. SMA was measured using axial slices at the level of the third lumbar vertebra. A threshold range of −30 to 150 HU was used to define muscle, and the SMA measurement included the abdominal, psoas, and paraspinal muscles. The skeletal muscle index (SMI) was calculated as SMA divided by height squared.

| CRP measurement
Serum concentrations of CRP were measured on PODs 1, 3, 5, and 7. The quantitative determination of CRP was carried out using an automated analytical system (N-Assay LA CRP-T Nittobo, Nittoubo Medical Co., LTD). CRP <0.15 mg/dL was considered normal.

| Definition of PICs
We defined PICs as either anastomotic leakage (extravasation of endoluminally administered water-soluble contrast agent on radiography), abdominal abscess formation (intra-abdominal pus collection confirmed by radiographic evidence or drainage), pancreatic fistula (drain output of any measurable volume of fluid on or after POD 3 with an amylase content greater than three times the serum amylase activity with medical management indicated [e.g. antibiotics/drainage]), 16 incisional surgical site infection (infection of the superficial and deep incisional surgical site with medical management), or pneumonia (infection of the lungs diagnosed by radiographic evidence and sputum culture). According to the Clavien-Dindo (CD) classification, 17 the severity of PICs was classified as grades 0-V, and patients with grade II or higher were defined as having PICs.

| Surgical procedure
All patients underwent laparoscopic-assisted distal gastrectomy or total gastrectomy (LATG) with D1, D1+, or D2 lymphadenectomy according to the Japanese Guidelines. After lymphadenectomy, a smalllaparotomy (<6 cm) was made in the upper abdomen for removal of the specimen and reconstruction. Omentectomy was performed in patients with sT3-T4. The omentum more than 3 cm away from the K E Y W O R D S C-reactive protein, gastrectomy, intra-abdominal fat, postoperative complications, stomach neoplasms gastroepiploic arcade and vessels of the omental branch was preserved in patients with sT1-T2. Reconstruction after laparoscopicassisted distal gastrectomy was performed by the Billroth I, Billroth II, or Roux-en Y approach, whereas Roux-en Y reconstruction was always used after LATG. After reconstruction, a closed drain system was placed in subhepatic area.

| Statistical analysis
All statistical analyses were performed using SPSS version 25 18 The optimal cut-off values were determined by maximizing Youden's index (defined as sensitivity + specificity − 1). 19 P < .05 was considered statistically significant.

| Patient characteristics
A total of 435 patients undergoing laparoscopic surgery for gastric cancer were identified. Twenty-one patients were excluded for unavailability of preoperative computed tomography images (19 patients)

| Clinicopathological findings stratified by VFA quartile
Patients were divided into four groups according to their VFA quartile ( Table 1). The proportion of women decreased as the VFA increased. Higher VFA was correlated with longer operation time and increased operative blood loss. The incidence of postoperative complications in the 4th VFA quartile of group was significantly higher than in those of the other groups.   the Table S1. The CRP level on POD3 increased in a stepwise manner with VFA increases in patients with and without PICs. In patients without PICs, the median CRP level within the fourth VFA quartile was significantly higher than that in the first and second, as were median levels in the third quartile.

| Effect of VFA on CRP on POD3 according to simple and multiple regression analysis
We examined the effect of factors that had been reported to be useful for predicting infectious complications after gastric cancer surgery on CRP levels on POD3 (Table 2)

| Diagnostic accuracy of CRP and WBC on POD1 and POD3 for the PICs prediction
To determine the most useful biochemical tests of the acute systemic inflammatory response to diagnose PICs, we performed the ROC analysis of PICs prediction using CRP and WBC value on POD1 optimal cut-off value of 11.8 mg/dL ( Figure 3B). Figure 3E shows the scatter plots of VFA and CRP levels on POD3 according to PIC status.

| Univariate and multivariate analysis for factors related to PICs
We examined clinicopathological factors related to the development of PIC after gastrectomy. Table 3 shows the results of univariate and multivariate analyses of clinicopathological factor for PICs prediction.  As shown in Table 4, the sensitivity for PICs in the 2nd, 3rd, and 4th

| D ISCUSS I ON
It is a known trend that the inflammatory response is high in obese patients, who show elevated serum levels of inflammatory cytokines, such as tumor necrosis factor-α, interleukin-6, and CRP. [20][21][22] However, there are few reports about differences in the inflammatory response change after surgery between obese patients and non-obese patients.   Note: Data are presented as mean ± SD or number.

TA B L E 4
Accuracy of CRP levels on POD3 in diagnosing postoperative infectious complication