Sepsis increases perioperative metastases in a murine model

Background Cancer surgery can promote tumour metastases and worsen prognosis, however, the effect of perioperative complications on metastatic disease remains unclear. In this study we sought to evaluate the effect of common perioperative complications including perioperative blood loss, hypothermia, and sepsis on tumour metastases in a murine model. Methods Prior to surgery, pulmonary metastases were established by intravenous challenge of CT26LacZ colon cancer cells in BALB/c mice. Surgical stress was generated through partial hepatectomy (PH) or left nephrectomy (LN). Sepsis was induced by puncturing the cecum to express stool into the abdomen. Hemorrhagic shock was induced by removal of 30% of total blood volume (i.e. stage 3 hemorrhage) via the saphenous vein. Hypothermia was induced by removing the heating apparatus during surgery and lowering core body temperatures to 30 °C. Lung tumour burden was quantified 3 days following surgery. Results Surgically stressed mice subjected to stage 3 hemorrhage or hypothermia did not show an additional increase in lung tumour burden. In contrast, surgically stressed mice subjected to intraoperative sepsis demonstrated an additional 2-fold increase in the number of tumour metastases. Furthermore, natural killer (NK) cell function, as assessed by YAC-1 tumour cell lysis, was significantly attenuated in surgically stressed mice subjected to intraoperative sepsis. Both NK cell-mediated cytotoxic function and lung tumour burden were improved with perioperative administration of polyI:C, which is a toll-like receptor (TLR)-3 ligand. Conclusions Perioperative sepsis alone, but not hemorrhage or hypothermia, enhances the prometastatic effect of surgery in murine models of cancer. Understanding the cellular mechanisms underlying perioperative immune suppression will facilitate the development of immunomodulation strategies that can attenuate metastatic disease.


Background
Severe trauma causes compensatory changes in immune, neural, endocrine, and metabolic function [1]. Likewise, surgical stress can lead to the onset of prothrombotic and immunosuppressive changes during the postoperative period [2,3]. Correlative studies have confirmed an association between postoperative complications, immune suppression, and worsened cancer prognosis [4][5][6][7]. Moreover, our group and others have proposed surgery-induced cellular immune suppression as a primary factor in the progression of cancer, including local recurrence and metastatic disease [8][9][10][11][12][13][14][15][16][17][18][19][20]. In humans, suppression of the cellular immune response following major surgery appears to peak at 3 days [21], but can also persist for weeks [17,22,23]. These immunosuppressive changes are characterized by an imbalance in plasma cytokine levels (i.e. a decrease in the levels of interleukin (IL)-2 [24] and IL-12 [25] and an increase in the levels of IL-6 [24,26,27] and IL-10 [28]) and a decrease in the number and function of circulating CD8 + T cells [29], dendritic cells [30], and natural killer (NK) cells [8,12,31]. Specifically, our group reported on the association between coagulation and NK cell function in the development of metastases following cancer surgery [8]; while, more recently, we employed validated murine models of surgical stress and spontaneous metastases [11] to provide in vivo evidence of global NK cell dysfunction in postoperative metastatic disease.
Modern surgical techniques minimize the adverse consequences of perioperative events, such as intraoperative blood loss, sepsis, and hypothermia. Despite this, however, severe intraoperative blood loss occurs in approximately 6-10% of patients with advanced cancer [32], while surgery accounts for 30% of all sepsis diagnoses in the US annually [33]. Furthermore, 8.5% of all cancer-related deaths are due to the concurrent onset of severe sepsis [34], and hypothermia, which is defined as a core body temperature of < 36°C, occurs in 70% of postoperative patients [35].
Clinical studies in cancer patients have confirmed an association between perioperative factors such as hypothermia [36], blood loss [37,38], and postoperative infections [39,40], and increased cancer recurrence and reduced cancer-specific survival following cancer surgery.
Despite the epidemiological evidence linking perioperative complications with increased surgical stress and worsened cancer outcomes, the role of intraoperative blood loss, sepsis, and hypothermia in immunosuppression and metastatic disease remains poorly understood. Our study incorporates three surgical murine models of colorectal cancer (CRC) to investigate the effect of blood loss, sepsis, and hypothermia on NK cell function and metastatic disease. Taking measures to reduce perioperative complications and/or employing preoperative neoadjuvant immunotherapy will help to improve survival outcomes and reduce cancer recurrence.

Animals
Female age-matched (6-8 weeks old at study initiation) BALB/c mice (Charles River Laboratories, Wilmington, MA, USA) were housed in specific pathogen-free conditions. The number of mice employed per experiment is indicated in the figure legends. Animal studies complied with the Canadian Council on Animal Care guidelines and were approved by the University of Ottawa Animal Research Ethics Board.

Induction of experimental metastasis and surgical stress
Mice were subjected to 2.5% isofluorane (Baxter Corporations, Mississauga, ON, CA) for the induction and maintenance of anesthesia. Routine perioperative care for mice, including the subcutaneous administration of buprenorphine (0.05 mg/kg) for pain control the day of surgery and every 8 h for 2 days following surgery, was conducted in concordance with University of Ottawa protocols. Surgical stress was induced via an abdominal laparotomy (i.e. 3-cm midline incision), which was preceded by an intravenous challenge with 3e5 CT26LacZ cells to establish pulmonary metastases. Abdominal laparotomy was commenced 10 min following tumor inoculation, as previously described [11]. Animals were euthanized at 18 h or 3 days following tumor inoculation and their lungs were stained with X-gal (Bioshop Canada Inc., Burlington, ON, CA), as described previously [41]. The total number of surface metastases on the largest lung lobe (left lobe) were quantified using a stereomicroscope (Leica Microsystems, Richmond Hill, ON, CA).

Hypovolemic stress model
Hypovolemia was induced by preoperatively bleeding mice prior to tumour inoculation. Mice were bled either 20% (300 uL) or 30% (450 uL) of their total blood volume by puncturing the saphenous vein just above the foot. Systolic arterial pressure (SAP) in conscious mice before and after saphenous vein bleeding was measured using a tail-cuff sphygmomanometer. Mice were kept in a warmed black box and an inflatable cuff was applied to the base of the tail. The tail of each mouse was then placed on a piezoelectric sensor for analysis of the pressure waveforms.

Hypothermia stress model
Intraoperative hypothermic shock was induced by placing mice directly on the metal surgical surface without a heating pad immediately following tumour inoculation. Mice were kept under hypothermic conditions and anesthesia for approximately 2 h and were subsequently housed under normothermic conditions. Rectal temperatures were recorded every 15 min throughout the procedure to verify that hypothermia was maintained.

Sepsis stress model
Intraoperative polymicrobial sepsis was induced in mice by cecal puncture at the time of abdominal laparotomy (i.e. 3-cm midline incision). Polymicrobial sepsis was confirmed by Gram stain of peritoneal lavage fluid, which was isolated 18 h following surgery. Bacterial counts were determined by serial dilution of peritoneal lavage fluid and overnight culture on tryptic soy broth agar plates at 37°C. We also investigated whether antibiotic treatment with Imipenem, which was administered intravenously at 0.5 mg, or treatment with poly(I:C), a toll-like receptor (TLR)-3 ligand, at 150 μg/ 200 μL PBS had an impact on lung tumour burden.

Ex-vivo NK cell cytotoxicity assay
Chromium-release assays were conducted as previously described [42]. Briefly, splenocytes were isolated from surgically stressed and control mice 18 h after surgery (n = 3 for each treatment group and each E:T ratio). Pooled and sorted NK cells were resuspended at a concentration of 2.5 × 10 6 cells/mL. These cells were then mixed with chromium-labeled YAC-1 target cells, which were resuspended at a concentration of 3 × 10 4 cells/mL at various effector-to-target (E:T) ratios (i.e. 50:1, 25:1, 12:1, and 6:1).

Statistical analysis
Statistical tests were performed using GraphPad Prism (GraphPad, San Diego, CA, USA). One-way ANOVAs, factorial ANOVAs with Tukey correction for multiple comparisons, and student's t-tests with equal variances were conducted. Data were reported as the mean ± standard error of the mean (SEM). An alpha value of < 0.05 was considered to be statistically significant.

Severe hypovolemia increases pulmonary metastases, but is not additive when combined with surgical stress
To discern the effect of hypovolemic shock on perioperative metastases, mice were first exsanguinated through the saphenous vein while systolic blood pressure was measured using a tail cuff sphygmomanometer (Fig. 1a). A significant reduction in tail cuff blood pressure was observed following the loss of 450 uL of blood, which is representative of 30% of the total blood volume of a 25-g mouse (71.67 ± 2.186 mmHg vs. 107.7 ± 1.453 mmHg, p = 0.0002); moreover, the reduction in tail cuff blood pressure was exacerbated when 450 uL of blood loss was combined with surgical stress (61.67 ± 1.667 mmHg vs. 107.7 ± 1.453 mmHg, p < 0.0001) (Fig. 1b). In the absence of surgical stress, hypovolemic changes due to 30% blood loss significantly increased the number of pulmonary metastases when compared to mice that did not undergo any blood loss (438.3 ± 56.89 vs. 205.0 ± 24.88, p = 0.0212) (Fig. 1c). In addition to hypovolemia, surgical stress alone significantly increased pulmonary metastases compared to mice that did no undergo surgery (287.5 ± 39.01 vs. 95.40 ± 32.35, p = 0.0065). However, a combination of 30% blood loss and surgical stress did not further increase the number of pulmonary metastases above surgical stress alone (285.6 ± 35.94 vs. 287.5 ± 39.01, p = 0.9725) (Fig. 1d).

Severe hypothermia does not increase pulmonary metastases
Next, we sought to assess the effect of perioperative hypothermic shock on metastases in a surgical murine model. Anaesthetized animals were maintained at either normothermic (35.22°C-37.95°C) or hypothermic (26.35°C-29.03°C) temperatures for a 3-h period following injection of the tumour cells and surgical stress ( Fig. 2a and b). We observed that hypothermia alone did not significantly increase pulmonary metastases when compared to normothermic mice (27.75 ± 7.667 vs. 21.38 ± 7.720, p = 0.5673) (Fig. 2c). Surgical stress in normothermic conditions did significantly increase the number of pulmonary metastases compared to mice that did not undergo surgery (391.5 ± 38.18 vs. 136.4 ± 21.47, p = 0.0024) (Fig. 2d). When combined with surgical stress, however, hypothermia did not further increase the number of pulmonary metastases when compared to mice that underwent surgical stress under normothermic conditions (391.5 ± 38.18 vs. 341.8 ± 80.90, p = 0.6265).
Perioperative polymicrobial sepsis increases pulmonary metastases compared to surgical stress alone Next, we assessed the effects of sepsis on pulmonary metastases by contaminating the peritoneal cavity with stool expressed through caecal puncture (Fig. 3a). Peritoneal lavage contained both Gram-negative and Grampositive bacteria, as well as coccus and bacillus-shaped bacteria, confirming that the caecal puncture lead to polymicrobial sepsis (Fig. 3b). Surgical stress alone resulted in a significant increase in pulmonary metastases when compared to control mice (374.3 ± 20.68 vs. 266.4 ± 16.15, p = 0.0006); furthermore, we observed that a combination of polymicrobial sepsis and surgical stress resulted in a significant increase in pulmonary metastases when compared to mice that underwent surgical stress in the absence of sepsis (480.3 ± 18.98 vs. 374.3 ± 20.68, p = 0.0010) (Fig. 3c). Previous studies have shown that suppression of NK cell cytotoxic activity is responsible for the increase in cancer burden following surgical stress. In this study, we demonstrate that sepsis, in conjunction with surgical stress, significantly attenuated NK cell cytotoxic activity below that of surgical stress alone (Fig. 3d).

Perioperative NK cell stimulation reduces metastases and restores NK cell function in the presence of sepsis
We have shown that postoperative immune dysfunction can be ameliorated through the use of NK cellstimulating agents such as poly(I:C), a double-stranded RNA mimetic. In this study, we examined whether sepsis-induced postoperative pulmonary metastases could be suppressed by perioperative immune stimulation with polyI:C (Fig. 4a). We demonstrated that treatment with polyI:C alone significantly decreased the number of pulmonary metastases compared to mice undergoing surgical stress and sepsis with no perioperative therapy (18.25 ± 8.390 vs. 175.0 ± 13.48, p = 0.0002); moreover, antibiotics alone did not affect the number of pulmonary metastases when compared to surgically stressed mice with sepsis that were not administered perioperative therapy (185.3 ± 19.38 vs. 175.0 ± 13.48, p = 0.6794) (Fig. 4a). Furthermore, antibiotic therapy did not further decrease the number of pulmonary metastases in mice undergoing surgical stress and sepsis in the presence of polyI:C (25.80 ± 4.306 vs. 18.25 ± 8.390, p = 0.4213). In support of a role for NK cells in mediating this effect we also demonstrated that the addition of polyI:C enhanced NK cell cytotoxic activity by 10-fold, a magnitude similar to the reduction in metastases seen with perioperative use (Fig. 4b).

Discussion
Perioperative complications, specifically infection, decrease long-term survival [5,43] and promote recurrence in patients with CRC [44]. Although hypovolemia in the absence of surgical stress did lead to an increase in pulmonary metastases, our findings demonstrate that neither severe intraoperative hypovolemia nor hypothermia impact the prometastatic effects of surgical stress. Correlative clinical studies confirm that postoperative infections following surgery can accelerate the time to cancer recurrence [45][46][47]. Here, using murine models we demonstrate that polymicrobial sepsis in conjunction with surgical a b c d Fig. 1 Hemorrhagic shock does not increase metastatic disease. a Experimental overview. BALB/c mice were bled through the saphenous vein (indicated by the white arrow) and subsequently injected intravenously (IV) through the tail vein with 3 × 10 5 CT26LacZ cells. Approximately 1 h later, surgical stress (sx) was generated by laparotomy (Lap) (5 cm incision). Mice were sacrificed at 72 h to quantify lung metastases. b Blood pressure is reduced following surgical stress and blood loss. Blood pressure (mmHg) was measured following a 5-day training period (Day 1-5), prior to bleeding (Pre), immediately following bleeding (Post-BL), and immediately following surgical stress (Post-Sx and BL, n = 3). c Blood loss increases metastatic burden. Lung metastases were measured on Day 3 following no blood loss (no BL, n = 3) or 20% (20% BL, n = 3) or 30% blood loss (30% BL, n = 4). d Blood loss does not increase metastatic disease in conjunction with surgical stress. Lung metastases were measure on Day 3 in mice that did not undergo surgical stress (No Sx, n = 5) and animals undergoing a laparotomy (Lap, n = 4) alone or in combination with 30% blood loss (Lap + 30% BL, n = 5). Error bars represent ± SEM stress facilitates the development of perioperative lung metastases. Our results suggest that the combined immunosuppressive effects of surgical trauma and sepsis dampen anti-tumour immune responses, ultimately leading to an increase in metastases. In addition to the immunosuppressive effects of surgical stress, severe sepsis can induce lymphocyte exhaustion [48], apoptosis of immune cells [49,50], and a predominance of immunoregulatory cells, including regulatory T cells [51,52] and myeloid-derived suppressor cells [53]. This highly suppressive environment likely worsens the already immunosuppressive environment present in most cancer patients in need of surgical intervention [54,55]. Thus, the immunosuppressive effects of surgery, sepsis, and cancer may interact to severely dampen immune activation and increase the likelihood of cancer recurrence and metastatic disease.
Our findings also suggest that sepsis induces its prometastatic effect by inhibiting NK cell cytotoxic function. In the cancer microenvironment, the anti-tumour function of NK cells is suppressed [56], while a decrease in NK cell number and function in patients undergoing surgery for CRC is associated with heightened mortality and cancer recurrence suggesting that the suppressive effects of sepsis likely exacerbate the already impaired NK cell function [57,58]. In agreement with our findings, previous studies have demonstrated that sepsis in a non-surgical context can impair NK cell cytotoxicity [59], a finding that has been attributed to a heightened activation of regulatory cell subsets [60]. In particular, murine sepsis models have shown that an increase in regulatory T cells contributes to post-sepsis immunosuppression and potentiates tumour growth [61]. so, based on our findings, we suggest that the inhibition of NK cell function is a key player in perioperative cancer recurrence following surgical stress and septic insult. Tumour-infiltrating NK cells and lymphocytes are associated with improved prognosis in several malignancies [62][63][64][65][66]. The enhancement of preoperative NK cell activation with PolyI:C, a TLR3 ligand, to counteract the immunosuppressive effects of surgery and sepsis and attenuate perioperative metastases formation is largely in agreement with the inhibitory effects of poly(I:C) upon tumour outgrowth in non-surgical models of lung metastases [67]. While polyI:C is ineffective in primates because of inactivation by natural enzymes, other NK stimulators, such as poly-ICLC [68] (stabilized with poly-lysine) or a virus-derived TLR agonist, like the influenza vaccine, could be safely and effectively employed in the perioperative period. Taken together, boosting NK cell activation may counteract the immunosuppressive effects of sepsis and protect against the development of metastatic disease and has potential as a perioperative cancer immunotherapeutic strategy.

Conclusions
In conclusion, our study is the first to utilize a murine model to investigate the effects of surgical complications on cancer recurrence in the perioperative period. Our findings demonstrate that intraoperative sepsis, but not intraoperative blood loss or hypothermia, contributes to the development of greater metastatic disease. We also demonstrate that perioperative sepsis-induced metastases are mediated by a suppression of NK cell cytotoxicity and can be reversed by TLR-mediated stimulation of NK cells. Further studies are required to determine whether enhancing NK cell function can prevent the development of perioperative metastatic disease in patients undergoing cancer surgery.