Secondary Logo

Journal Logo

Basic Science Aspects


Hugunin, Kelly M.S.; Fry, Christopher; Shuster, Katherine; Nemzek, Jean A.

Author Information
doi: 10.1097/SHK.0b013e3181cdc412



Sepsis research relies heavily on animal models to recreate a complex syndrome. In general, the models that most closely resemble clinical sepsis in humans, such as cecal ligation and puncture (CLP), require abdominal surgery to create a specific focus of infection that induces a systemic response (1-3). The use of surgery to induce peritonitis in these models sparks concern in animal use committees (4). When considering approval of federally funded animal use, Institutional Animal Care and Use Committees (IACUC) must interpret the Public Health Service Policy on Humane Care and Use of Laboratory Animals, which states that "procedures that may cause more than momentary or slight pain or distress to animals will be performed with appropriate sedation, analgesia, or anesthesia, unless the procedure is justified for scientific reasons in writing by the investigator" (5). Consequently, investigators are charged to either administer analgesics to rodents after CLP or provide adequate justification for withholding the drugs.

There are several reasons why investigators may not wish to use analgesics in sepsis models. One question is whether animals with sepsis actually experience or are aware of pain. Human patients develop sepsis-associated encephalopathy (SAE), a diffuse cerebral dysfunction induced by a systemic response to infection, which has been identified at all stages of sepsis (6). As a result of multiple organ failure or other unidentified causes, humans with SAE have attenuated or even absent pain responses (6). This characteristic may also be present in animal models, although the role of SAE has not been formally investigated as a determinant for analgesic use. Of greatest concern to investigators, the use of analgesics has the potential to alter the course of the disease syndrome. Because sepsis is caused by aberrant inflammatory responses, the use of steroids and nonsteroidal anti-inflammatory agents is problematic. Opioids may also have immunomodulatory effects. For instance, exogenous morphine administration inhibits various antibody and cellular responses, including natural killer (NK) cell activity, cytokine expression, chemokine-induced chemotaxis, and phagocytic activity (7, 8). In fact, morphine administration can increase susceptibility to numerous pathogens including viruses, bacteria, and parasites (9). In addition to these considerations, the use of analgesics in the research setting may involve extra time, money, and personnel to administer treatments. The use of controlled substances increases the need for security and accurate reporting. For these reasons, there is a general lack of analgesic use in sepsis models.

Despite the previously mentioned justifications, the omission of analgesics from animal protocols using the CLP model is controversial for many reasons. In humans, analgesics may be used when sepsis occurs secondary to conditions such as trauma, burns, and surgery (7, 10, 11). In addition, unrelieved pain creates many effects that can interfere with metabolic and immunologic functions. Combined, these two considerations suggest that modeling sepsis without analgesics provides an inaccurate clinical picture in some situations. There is also concern that the innate protective behaviors of prey animals, including rodents, will mask obvious signs of pain. In addition, it is not known if an unresponsive animal is aware of pain (12), particularly because the existence of SAE has not been established in sepsis models. Therefore, the extra investment of time and money to administer analgesics will not outweigh the benefits of pain management in the eyes of many IACUCs trained to consider "that procedures that cause pain or distress in human beings may cause pain or distress in other animals" (13). The aforementioned reasons fuel increasing demands from Association for Assessment and Accreditation of Laboratory Animal Care International and welfare groups, ensuring that pain management issues in research studies will not go away.

Currently, there is a lack of studies addressing the effects of analgesics in sepsis models. Consequently, neither investigators nor IACUCs have direct evidence to justify their positions on analgesic use. Therefore, the purpose of this study was to evaluate the use of analgesic agents in a model of CLP-induced polymicrobial sepsis. Our study examined buprenorphine, a commonly used analgesic in laboratory rodents. In addition, we examined tramadol, which may prove a more convenient option for analgesia in the research setting. Tramadol is a noncontrolled opioid analgesic used frequently in veterinary species in the United States and in the human population in Europe. It is reported to have minimal effects on immune parameters compared with morphine (14). We sought to determine the effects of these opioids on mortality and inflammation including cell counts and cytokine levels in local (peritoneum), systemic (plasma), and distant (lung) compartments.


Study design

Mice underwent either cecal ligation and puncture (CLP) or sham laparotomy (SHAM) then were randomly assigned to one of four treatment groups: (B) buprenorphine HCl; (Th) high-dose tramadol HCl; (Tl) low-dose tramadol; or (N) negative control. For a mortality study, mice were observed for 3 weeks after undergoing CLP. To examine the effect of analgesics on inflammation, mice were killed at 12, 24, or 48 h postsurgery (n = 12 mice per treatment per time point per surgery). The experimental reproducibility was assured by performing CLP in several small independent trials (n = 12-16 mice per trial) and then combining the results for this study. Blood collection, peritoneal lavage, and bronchoalveolar lavage (BAL) were performed at harvest. White blood cell (WBC) counts, differentials, and cytokine levels (IL-6, IL-1β, TNF-α, and IL-10) were determined from these biological samples.

Experimental animals

Female outbred Hsd:ICR (ICR) mice weighing 21 to 25 g were obtained from Harlan Laboratories (Indianapolis, Ind). Mice were housed four to five per cage in static microisolation cages in a specific pathogen-free barrier facility. Mice had ad libitum access to food (Laboratory Rodent Diet 5001; PMI LabDiet, St Louis, Mo) and water. The animal housing room was maintained on a 12:12-h light/dark cycle with constant temperature (72°F ± 2°F). Mice were acclimated for at least 7 days before experimental use. All procedures were approved by the University of Michigan's Animal Care and Use Committee. The experiments were performed in the adherence to the National Institutes of Health guidelines on the use of laboratory animals.

Surgical procedures

All mice underwent either CLP or SHAM surgery. For these procedures, mice were maintained on isoflurane. For CLP, a single ligature using 4-0 silk was tied immediately distal to the ileocecal valve, and two punctures were made with a 21-gauge needle. The cecum was gently expressed to ensure patency of punctures (15). The abdominal musculature was closed with sutures and followed by closure of the skin with tissue glue. This CLP model leads to approximately 40% to 50% mortality within 5 days. Mice receiving SHAM underwent identical procedures to CLP without ligation and puncture of the cecum.

Analgesic dosing

All analgesic treatments were provided subcutaneously in 1.0 mL Ringer's lactate solution (LRS) immediately after induction of anesthesia and again at 10- to 14-hour intervals for up to four total doses. The B group was given 0.1 mg/kg buprenorphine (Bedford Labs, Bedford, Ohio) based on the high end of standard reference ranges for mice (16). Tramadol (BioChemika puriss, Sigma-Aldrich) was given at either 80 mg/kg for the Th group or 20 mg/kg for the Tl group based on doses provided in studies of its effects on immune parameters (17, 18) and antinociceptive properties (14). The N group received an equal volume of LRS without analgesics.

Tramadol HCl preparation

Tramadol is not available in an injectable form in the United States but can be purchased as a nonpharmaceutical grade powder. To prepare it for injection, the powdered tramadol was dissolved in sterile saline and passed through a 0.22-μM filter into a sterile glass vial. Based on stability data, tramadol HCl is stable in solution for at least 15 days (19). For our experiments, solutions were kept for up to 8 days after preparation but were generally used within 2 days of preparation. Limulus amebocyte lysate (Cambrex Bio Science Inc, Walkersville, Md) assay for endotoxin, as well as bacterial and fungal cultures, was negative for an aliquot of the filtered solution.

Blood collection and processing

Mice were anesthetized with a subcutaneous injection of ketamine HCl (87 mg/kg; Ketaset, Fort Dodge Laboratories, Fort Dodge, Iowa) and xylazine (13 mg/kg; Anased, Lloyd Laboratories, Shenandoah, Iowa), and 450 μL of blood was collected from the retro-orbital sinus into 2-mL tubes containing 50 μL of 169 mmol EDTA. Blood was centrifuged (2,000g, 5 min), and the plasma was stored at −20°C for later cytokine analysis. An additional 50 μL blood was collected into purple top microtainer tubes with EDTA (BD Microtainer tubes with EDTA; BD Laboratories, Franklin Lakes, NJ) for a complete blood count using a Hemavet Mascot Multispecies Hematology System Counter (CDC Technologies, Oxford, Conn). Mice were then euthanized by cervical dislocation.

Peritoneal lavage

After euthanasia, 10 mL of Hanks balanced salt solution (Invitrogen, Grand Island, NY) containing 1:100 heparin sodium (1,000 USP units/mL; Abraxis, Schaumberg, Ill) was injected into the mouse peritoneal cavity, and 8 mL of the solution was retrieved using a 21-gauge needle. The peritoneal lavage fluid was centrifuged (600g, 5 min), and the supernatant saved at −20°C for later cytokine analysis. The cell pellet was reconstituted in 200 μL of RPMI Medium 1640 (Invitrogen, Grand Island, NY) with 0.1% heat-inactivated fetal bovine serum (Invitrogen, Grand Island, NY) and used for cell count and differentials.

Bronchoalveolar lavage

After peritoneal lavage, 0.3 mL aliquots of Hanks balanced salt solution was introduced into the mouse lungs via the trachea and aspirated. The collected fluid was partitioned into two 1-mL aliquots, centrifuged (600g, 10 min), and 1 mL of supernatant was saved at −20°C for later cytokine analysis. The two cell pellets were combined in 200 μL of RPMI and processed for cell counts and differentials.

Cell counts and differentials

The reconstituted cell pellets from the peritoneal lavages and BALs were counted using a Coulter Counter model Z1 after red blood cell lysis with Zap-Oglobin II (Coulter Corp, Miami, Fla). Slides were loaded with 1 × 105 cells, centrifuged (500 RPM, 5 min), and stained with Diff-Quick (Baxter, Detroit, Mich). Differentials (300 cells) were counted under light microscopy.

Cytokine ELISA

Cytokines were measured in plasma (1:10 dilution), as well as peritoneal and lung lavage fluids (1:2 dilution) using sandwich enzyme-linked immunosorbent assay (ELISAs). Matched pairs (biotinylated and nonbiotinylated) of antimurine antibodies against IL-1β, IL-6, TNF-α, and IL-10 with their recombinant proteins (R&D Systems, Minneapolis, Minn) were used in methods previously described by this laboratory (20). Peroxidase-conjugated streptavidin (Jackson ImmunoResearch Laboratories) and the color reagent TMB were used as the detection system. The reaction was stopped with 1.5N sulfuric acid, and the absorbance was read at 465 and 590 nm.

NK cell study

Treatment groups were repeated in animals with CLP, and mice were euthanized at 48 h after surgery. Spleens were removed and processed for flow cytometry.

Flow cytometry

Whole spleens from individual animals were teased apart in 10 mL of phosphate-buffered solution (PBS). The resulting cell suspension was filtered using a 100-μm filter (BD Falcon, San Jose, Calif) into individual 15-mL tubes. Cells were centrifuged and resuspended in 5 mL red blood cell lysis buffer (eBiosciences, San Diego, Calif). Cells were centrifuged and resuspended in 1× PBS and counted using a hemacytometer (Hausser Scientific, Horsham, Pa). Trypan blue (Invitrogen, Carlsbad, Calif) was used to demonstrate viability of greater than or equal to 95% in suspensions. After total splenic cell counts, cells were resuspended in PBS with 0.1% sodium azide and 1.0% bovine calf serum (HyClone, Logan, Utah) at 1.0 × 106 cells/mL. Cells were then incubated for 5 minutes (4°C) with 0.5 μg FcγII/III reagent (BD Pharmingen) to block Fc receptors. The cells were incubated for 30 min at 4°C with 2.5 μg/mL of each of the following fluorescently labeled antibodies: hamster antimouse CD3ε APC-Cy7 (BD Pharmingen), antimouse NK1.1 APC (eBiosciences), and antimouse CD69 PE (eBiosciences). Control antibodies were used at 2.5 μg/mL for each experimental treatment and consisted of APC-Cy7 hamster IgG1 (BD Pharmingen), APC mouse IgM (BD Pharmingen), and PE hamster IgG1 (BD Pharmingen). The fluorescence was measured with a BD LSR II flow cytometer. Compensation was performed using WinList for 32 software (Verity Software House, Inc, Topsham, Maine).

Statistical analysis

For the mortality study, 3-week mortality rates were calculated for each treatment group. Pairwise Fisher exact tests were used to determine significant differences in mortality rates between treatment groups. Relative risks (RRs) and corresponding two-sided 95% confidence intervals (95% CIs) were calculated to estimate the risk of death in one treatment group relative to another. In addition, Kaplan-Meier survival curves were calculated for each treatment group, and differences between group survival were analyzed by log-rank tests.

For the immune parameters, data were analyzed for normality using D'Agostino-Pearson omnibus normality tests. Parametric or nonparametric analyses were performed based on the distribution of the data. Treatment group means and SEs or medians and interquartile ranges (IQRs) were calculated for each parameter of interest for each type of surgery and time point. Within surgery type at each time point, the association between the parameter of interest and treatment was assessed using one-way ANOVA or Kruskal-Wallis tests. In addition, pairwise comparisons between treatment groups were made using Student t tests or Wilcoxon-Mann-Whitney U tests when the P value of the corresponding ANOVA or Kruskal-Wallis test for a particular parameter (within surgery and time point) was less than 0.25. This sequential analysis was performed to limit the number of pairwise comparisons performed because type I error rate for multiple comparisons was not controlled to maximize power to detect significant differences.


Mortality study

For this CLP model, there was 50% mortality for negative-control mice. Most of these mice died within the first 3 days after CLP, and no deaths occurred past 1 week post-CLP (Fig. 1). The Tl group most closely replicated this state, with 50% 3-week mortality and the most deaths occurring within the first week after CLP. Treatment with B resulted in slightly lower mortality (33%) during a similar time frame. The Th treatment was associated with a higher mortality rate (83%) and several late deaths that occurred after 2 weeks post-CLP. Compared with negative-control mice, death was 1.7 times as likely to occur in mice receiving Th (RR, 1.67; 95% CI, 0.90-3.10) but only 0.67 times as likely in mice receiving B (RR, 0.67; 95% CI, 0.25-1.78). Despite some variations among treatment groups, no analgesic treatment had a mortality rate significantly different from the negative control, and log-rank tests indicated no significant difference in survival for any treatment relative to the negative control. However, differences were evident between analgesic groups. The mortality rate for Th-treated mice was significantly higher than for B-treated mice (P = 0.0361). In addition, the risk of death for mice receiving Th was significantly greater compared with those receiving B (RR, 2.5; 95% CI, 1.08-5.79). Similarly, there was a significant difference in survival curves for mice receiving B compared with Th (Fig. 1).

Fig. 1:
Kaplan-Meier survival curves by opioid treatment. Mice underwent CLP followed by treatment with LRS, buprenorphine, low-dose tramadol, or high-dose tramadol. There were no significant differences between any analgesic treatment and the LRS controls. However, survival was significantly lower in the high-dose tramadol group when compared with that of the buprenorphine group. n = 12 per group. *Significant difference between buprenorphine and high-dose tramadol (log-rank analysis, P = 0.0176).

Immune parameter study

Cell counts and differentials


In negative-control SHAM mice, cell counts were within reference ranges for mice, although there was fluctuation over time (see Figure, Supplemental Digital Content 1, Cell counts decreased from 12 to 24 h post-SHAM and then rebounded from 24 to 48 h post-SHAM. The median counts at 12, 24, and 48 h for total WBCs (9,650/μL, 4,910/μL, and 6,750/μL, respectively), neutrophils (1,680/μL, 820/μL, and 1,475/μL, respectively), and lymphocytes (7,185/μL, 3,560/μL, and 5,220/μL, respectively) demonstrated this pattern, although median monocyte counts (295/μL, 190/μL, 205/μL, respectively) showed little variation. There were no significant differences between negative control and any treatment group for any peripheral WBC parameter in SHAM mice. However, at 48 h, SHAM mice treated with Th had significantly lower levels of blood neutrophils (median, 1,240/μL; IQR, 918-1,660/μL) compared with those treated with B (median, 1,710/μL; IQR, 1,403-2,015/μL; P = 0.0349).

Negative-control CLP mice showed typical patterns of decreased peripheral WBC counts as compared with SHAM mice. Within the CLP mice, several significant differences were detected between negative control and analgesic treatments (Fig. 2). At 12 h post-CLP, mice receiving B and Th had peripheral blood counts for all WBC types trending lower than for negative-control mice (Fig. 2). In fact, mice receiving B (mean, 900/μL; SE, 102/μL) had significantly lower blood neutrophils compared with negative-control mice (mean, 1,463/μL; SE, 222/μL; P = 0.0309). By 24 h, the neutrophil values were similar for negative-control and B- and Th-treated mice; however, mice receiving Tl (median, 760/μL; IQR, 510-972/μL) had significantly higher neutrophil levels compared with both negative-control mice (median, 410/μL; IQR, 330-668/μL) and B-treated mice (median, 450/μL; IQR, 370-690/μL; P = 0.0166 and P = 0.0455, respectively). At 48 h post-CLP, mice receiving Th had significantly fewer blood monocytes (mean, 47/μL; SE, 9/μL) compared with negative-control mice (mean, 80/μL; SE, 11/μL; P = 0.0287). Although differences between negative-control and treatment groups did occur, it is important to note that no treatment was involved in more than one significant difference relative to negative control and that the differences did not persist during the entire time course of the study.

Fig. 2:
Peripheral WBC counts by treatment over time for CLP mice. After CLP, mice were treated with either LRS or one of three analgesic regimens. Whole blood was collected for automated cell counts at various time points after CLP. Box indicates IQR. Horizontal line and cross represent the median and mean, respectively. Whiskers represent the smaller of either 1.5*IQR or range of the data. Dots represent outliers. For significance levels below 0.05, P values are presented for pairwise treatment comparisons. n = 12 per group.

In addition to differences involving negative control, there were also differences seen across analgesic treatments in CLP mice. At 24 h post-CLP, blood monocyte counts for mice receiving Th (median, 55/μL; IQR, 50-92/μL) were significantly lower compared with both B (median, 120/μL; IQR, 90-160/μL) and Tl-treated mice (median, 95/μL; IQR, 72-178/μL; P = 0.0126 and P = 0.0482, respectively). At 48 h post-CLP, mice receiving Th had the lowest counts for all cell types except neutrophils. The Th total WBC levels (mean, 1,855/μL; SE, 171/μL) were significantly lower than for B (mean, 3,091/μL; SE, 369/μL; P = 0.0065), as were monocytes (Th: mean, 47/μL; SE, 9/μL; B: mean, 95/μL; SE, 11/μL; P = 0.0028) and lymphocytes (Th: median, 680/μL; IQR, 570-1,120/μL; B: median, 1,390/μL; IQR, 1,080-2,170/μL; P = 0.0031). In addition, at 48 h post-CLP, mice receiving Tl (mean, 2,204/μL; SE, 193/μL) had significantly lower total WBC counts compared with those receiving B (P = 0.0457).


In negative-control SHAM mice, peritoneal cell counts demonstrated little variation from 12 to 24 h post-SHAM then increased at 48 h (see Figure, Supplemental Digital Content 2, These trends in median counts at 12, 24, and 48 h were evident for total cells (3.095, 4.175, and 6.371 × 106/mouse, respectively), macrophages (1.05, 1.65, and 3.07 × 106/mouse, respectively), and lymphocytes (0.11, 0.14, and 0.69 × 106/mouse, respectively), whereas there was little variation of peritoneal neutrophils (1.03, 1.91, and 1.26 × 106/mouse, respectively). When SHAM animals were treated with analgesics, the only significant differences occurred at 12 h when B-treated mice had significantly higher levels of total cell counts (P = 0.0194) and neutrophils (P = 0.0367) compared with negative-control mice.

For CLP mice, peritoneal cell counts were higher relative to SHAM mice at 12 h postsurgery and decreased over time (Fig. 3). There were no significant differences for any treatment relative to negative control in CLP mice. At 24 h postsurgery, mice receiving Tl (mean, 3.59 × 106/mouse; SE, 0.727 × 106/mouse) had significantly increased levels of peritoneal neutrophils relative to mice receiving B (mean, 1.58 × 106/mouse; SE, 0.276 × 106/mouse; P = 0.0145). Similar nonsignificant increases of peritoneal total cell and neutrophil counts were seen for mice receiving Tl at 12 and 24 h.

Fig. 3:
Peritoneal WBC counts by treatment over time for CLP mice. After CLP, mice were treated with either LRS or one of three analgesic regimens. Peritoneal lavage was performed for automated cell counts and manual differential at various time points after CLP. Box indicates IQR. Horizontal line and cross represent the median and mean, respectively. Whiskers represent the smaller of either 1.5*IQR or range of the data. Dots represent outliers. For significance levels below 0.05, P values are presented for pairwise treatment comparisons. n = 12 per group.

Macrophages accounted for greater than 98% of the total cell population in BAL fluid for all mice. For negative-control SHAM mice, BAL fluid median macrophage counts were similar at 12 and 24 h post-SHAM, then increased at 48 h post-SHAM (0.71, 0.51, and 1.15 × 105/mouse, respectively). There were no significant treatment differences in BAL fluid macrophages at any time point (see Figure, Supplemental Digital Content 3,

For negative-control CLP, BAL fluid macrophage counts were lower relative to SHAM at every time point (Fig. 4). Treatment differences involving negative control were seen at both 12 and 48 h post-CLP. At 12 h post-CLP, mice receiving B or Tl had significantly greater BAL fluid macrophage counts compared with negative controls (P = 0.0289 and P = 0.0117, respectively). At 48 h post-CLP, mice receiving Th (mean, 1.34 × 105; SE, 0.179 × 105) had significantly higher BAL fluid macrophage counts than negative-control mice (mean, 0.81 × 105; SE, 0.143 × 105; P = 0.0425).

Fig. 4:
Airway macrophage counts by treatment over time for CLP mice. After CLP, mice were treated with either LRS or one of three analgesic regimens. BAL was performed for automated cell counts and manual differential at various time points after CLP. Box indicates IQR. Horizontal line and cross represent the median and mean, respectively. Whiskers represent the smaller of either 1.5*IQR or range of the data. Dots represent outliers. For significance levels below 0.05, P values are presented for pairwise treatment comparisons. n = 12 per group.



Most negative-control SHAM mice had plasma cytokine levels below the lower limit of detection (LLD) for cytokine assays (see Figure, Supplemental Digital Content 4, For negative-control SHAM mice, the highest median levels of plasma IL-1β (105.6 pg/mL) and IL-10 (102.7 pg/mL) were seen at 24 h post-SHAM, whereas the highest median levels of IL-6 (111.6 pg/mL) were seen at 48 h. Median TNF-α levels (62.77 pg/mL) in negative-control SHAM mice did not vary over time. For plasma cytokines, only one significant difference relative to negative control was seen. At 24 h post-SHAM, plasma IL-1β was significantly increased in mice receiving Th (median, 220.7 pg/mL; IQR, 150.7-320.3 pg/mL) compared with negative controls (median, 105.6 pg/mL; IQR, 95.03-180.2 pg/mL; P = 0.0126). This result was also significant compared with mice receiving B (median, 95.03 pg/mL; IQR, 95.03-95.03 pg/mL; P < 0.0001).

Plasma cytokine levels for CLP mice not receiving treatment were higher compared with SHAM mice for all cytokines (Fig. 5). For CLP mice, there were no significant effects of treatment for any cytokine at any time point.

Fig. 5:
Plasma cytokines by treatment over time for CLP mice. After CLP, mice were treated with either LRS or one of three analgesic regimens. Blood was collected at various time points after CLP and plasma was analyzed for proinflammatory and anti-inflammatory cytokines by capture ELISA. Horizontal bars represent medians. For significance levels below 0.05, P values are presented for pairwise treatment comparisons. n = 12 per group.

For negative-control SHAM mice, peritoneal levels of cytokines were consistently below the LLD for most cytokines at all time points (see Figure, Supplemental Digital Content 5, However, at 24 h, TNF-α for mice receiving B (median, 59.6 pg/mL; IQR, 41.5-101.4 pg/mL) was significantly higher than for negative-control mice (median, 21.4 pg/mL; IQR, 21.4-54.0 pg/mL) and for those receiving Th (median, 21.4 pg/mL; IQR, 21.4-53.5 pg/mL; P = 0.0295 and P = 0.0081, respectively).

For CLP mice, peritoneal cytokines were increased compared with SHAM mice for all cytokines (Fig. 6). There was a single significant treatment effect compared with negative control for peritoneal cytokines in CLP mice. At 12 h, mice receiving Tl had significantly higher levels of peritoneal IL-10 (median, 83.9 pg/mL; IQR, 43.2 pg/mL, 131.5 pg/mL) compared with mice not receiving treatment (median, 26.8 pg/mL; IQR, 10.9-77.2 pg/mL; P = 0.0373). There were three significant differences across treatment groups for peritoneal cytokines in CLP mice. At 24 h, mice receiving B had significantly greater levels of peritoneal TNF-α (mean, 48.92 pg/mL; SE, 5.731 pg/mL) compared with mice receiving Tl (mean, 32.11 pg/mL; SE, 3.442 pg/mL; P = 0.0228), but significantly lower levels of IL-6 (median, 210.6 pg/mL; IQR, 76.85-974.1 pg/mL) compared with Tl (median, 807.0 pg/mL; IQR, 316.5-1,648 pg/mL; P = 0.0387). At 48 h, peritoneal levels of IL-10 were significantly greater for mice receiving Th (median, 9.565 pg/mL; IQR, 9.565-39.68 pg/mL) than for B-treated mice (median, 9.565 pg/mL; IQR, 9.565- 9.565 pg/mL; P = 0.0447).

Fig. 6:
Peritoneal cytokines by treatment over time for CLP mice. After CLP, mice were treated with either LRS or one of three analgesic regimens. Supernatant from peritoneal lavage was collected at various time points after CLP and analyzed for cytokines by capture ELISA. Horizontal bars represent medians. For significance levels below 0.05, P values are presented for pairwise treatment comparisons. n = 12 per group.

Values of BAL fluid cytokines in negative-control SHAM mice were consistently below the LLD for all cytokines at all time points except for IL-1β (median, 145.8, 171.5, 147.5 pg/mL for 12, 24, and 48 h, respectively). Compared with negative-control mice, there were two significant treatment effects seen for BAL fluid cytokines in SHAM mice (see Figure, Supplemental Digital Content 6, At 12 h, SHAM mice treated with Th had significantly higher levels of BAL IL-1β compared with negative controls (P = 0.0304). At 24 h, TNF-α levels were significantly higher in Th-treated mice (median, 43.06 pg/mL; IQR, 23.69-51.23 pg/mL) compared with negative controls (median, 10.00 pg/mL; IQR, 10.00-41.20 pg/mL; P = 0.0295) and B-treated mice (median, 10.00 pg/mL; IQR, 10.00-10.00 pg/mL; P < 0.0001).

Similar to SHAM mice, most negative-control CLP mice had BAL fluid cytokine levels below assay LLDs for all cytokines and time points except IL-1β (see Figure, Supplemental Digital Content 6, There were no significant treatment effects compared with negative control. There was one significant result seen across treatments for BAL fluid cytokines in CLP mice. At 24 h, IL-10 was increased in mice receiving Tl (median, 15.07 pg/mL; IQR, 7.713-83.37 pg/mL) compared with those receiving B (median, 7.71 pg/mL; IQR, 7.71-11.21 pg/mL; P = 0.0374) (see Figures, Supplemental Digital Content 7 and 8, and

NK cell study

At 48 hours after CLP, total splenocyte counts for mice receiving Th (mean, 4.2 × 107/mouse; SE, 0.36 × 107/mouse) were significantly increased compared with negative-control mice (mean, 3.1 × 107/mouse; SE, 0.25 × 107/mouse; P = 0.0415). Mean total splenocyte counts for B (mean, 2.8 × 107/mouse; SE, 0.52 × 107/mouse) and Tl (mean, 2.6 × 107/mouse; SE, 0.29 × 107/mouse) mice were similar to negative-control mice. Natural killer cells, identified as CD3 negative and NK1.1 positive, accounted for 5.9%, 6.0%, 8.6%, and 7.0% of the total splenocyte population for negative-control, B, Th, and Tl mice, respectively (Fig. 7A). The percentage of splenocytes represented by NK cells was significantly higher for Th mice compared with N (P = 0.0058) and B (P = 0.0023) mice. Similarly, total NK cell counts were significantly elevated in Th mice (mean, 3.6 × 106/mouse; SE, 0.32 × 106/mouse) compared with negative-control mice (mean, 1.8 × 106/mouse; SE, 0.32 × 106/mouse; P = 0.0063), as well as B (mean, 1.7 × 106/mouse; SE, 0.35 × 106/mouse; P = 0.0081) and Tl (mean, 1.8 × 106/mouse; SE, 0.31 × 106/mouse; P = 0.0050) mice (Fig. 7B). Gated on the NK cell population, the cells expressing CD69, a marker of activation, were identified and represented 80.4%, 76.3%, 82.3%, and 72.9% of the total NK cell population for negative-control, B, Th, and Tl mice, respectively (Fig. 7C). From the total NK cell counts, estimated total counts for NK cells expressing the activation marker were calculated. Total NK cells expressing the activation marker were significantly elevated in Th mice (mean, 2.9 × 106/mouse; SE, 0.30 × 106/mouse) compared with negative-control mice (mean, 1.4 × 106/mouse; SE, 0.23 × 106/mouse; P = 0.0062), as well as B (mean, 1.3 × 106/mouse; SE, 0.24 × 106/mouse; P = 0.0055) and Tl (mean, 1.3 × 106/mouse; SE, 0.19 × 106/mouse; P = 0.0021) mice (Fig. 7D).

Fig. 7:
Natural killer cell study 48 h post-CLP. Total NK cells, total CD69+ NK cells, and %CD69+ NK cells by treatment at 48 h post-CLP. After CLP, mice were treated with either LRS or one of three analgesic regimens. Spleens were harvested 48 h after CLP and analyzed for the presence of NK and activated NK cell markers. Figure (A) provides flow cytometry results for one representative mouse per group. The percentage of splenic NK cells, CD3(−) and NK1.1(+) cells, is provided in the lower left quadrant of each picture. Figures B to C represent total splenic NK cell counts, % of NK cells positive for activation marker CD69, and total activated NK cell counts. Horizontal bars represent means. **Statistical significance P < 0.01. n = 4 to 5 per group.


Morphine is the prototypical opioid and the source of many generalizations about the use of opioids. The well-documented mechanisms of morphine-induced immunosuppression include direct effects (opioid receptor binding on immune cells) and indirect effects (opioid activation of the hypothalamic-pituitary-adrenal axis and the sympathetic nervous system) (7, 8, 14). Although it is clear that morphine administration leads to immunosuppressive effects, other opioids may not. Opioids have been classified into immunosuppressive (morphine, codeine, methadone, remifentanil, fentanyl) and less-immunosuppressive (buprenorphine, hydromorphone, oxycodone, tramadol) categories (14). For our studies, we chose to examine the effects of two less-immunosuppressive opioids on a CLP model.

Buprenorphine, a thebaine-derived synthetic opiate, has partial agonist activity at the μ receptor (14). Studies have shown that buprenorphine has minimal effects on the immune system compared with other opioids (14, 21). In those preclinical studies in rodents, morphine and fentanyl produced immunosuppressive effects (e.g., reduction in lymphoproliferation [21], NK-lymphocyte activity [14, 21], T-cell or macrophage function [14], and splenic IL-2 and IFN-γ production [21]) that were not seen with buprenorphine administration. In contrast, other studies in rodents have indicated some pronounced effects on immunity after buprenorphine administration, including suppression of splenic NK cell activity, lymphocyte proliferation, and IFN-γ release (22), as well as decreased production of TNF-α (23). In addition, evidence suggests that buprenorphine can ameliorate the stress-induced neuroendocrine and immune system alterations (e.g., increased corticosterone and suppression of NK cell activity) associated with surgery (24).

In our study, treatment of mice with buprenorphine led to similar results as for negative-control mice for both CLP and SHAM. Out of 63 statistical comparisons of cell counts and cytokines made in CLP mice, two resulted in statistically significant differences between buprenorphine-treated mice and negative controls. For SHAM mice, there were three significant comparisons between buprenorphine-treated mice and negative controls. Of these five significant differences, four occurred at the 12-h time point. Significant differences observed in CLP mice were not consistent with those seen in SHAM mice. In fact, at 12 h, peripheral neutrophil counts relative to negative controls were significantly lower in CLP mice but significantly greater in SHAM mice receiving buprenorphine. Mortality during 3 weeks was not significantly different for buprenorphine mice compared with negative controls, a result that was not surprising given the overall lack of consistent differences observed in the measured immune parameters.

Our studies also evaluated tramadol, a centrally acting analgesic with low affinity for μ-opioid receptors (17). Tramadol also induces antinociception by inhibiting neuronal noradrenaline and serotonin reuptake, as well as stimulating serotonin release (17). There are a few published studies on the effects of tramadol on serum cytokines. One study determined that tramadol caused no differences in serum levels of IL-6 and IL-10 in human patients after pulmonary lobectomy compared with saline controls (25). Several studies suggest that tramadol has immune-stimulatory effects that reverse the suppressive effects of surgery on IL-2 production (17, 25), NK cell activity (17, 26), splenocyte proliferation (17), and T-lymphocyte proliferation (14). In a rat model of CLP, tramadol did not change gastrointestinal transit, whereas fentanyl significantly decreased transit (27). This study did not investigate the effects of tramadol on immune parameters.

Similar to our findings for buprenorphine, only a few significant differences were seen comparing tramadol-treated mice with negative controls. In CLP mice, there were two and three significant differences compared with negative control for high- and low-dose tramadol-treated mice, respectively. Mortality in the tramadol high group was 33% higher and deaths were seen later compared with negative controls. Although at 48 h post-CLP, peripheral cell counts were lower in the high-dose tramadol-treated group, the results of early cell counts and cytokines did not explain the increased mortality rates; other factors may have contributed. We examined splenic NK cells at 48 h post-CLP across all groups. These data showed significant increases for most of the parameters examined (total splenocytes, % NK cells, NK cell totals, and total activated NK cells) in Th mice compared with all other groups. These data are consistent with previous reports of increased T-cell proliferation and NK cell activity (14, 17, 26). This may partially explain the increased mortality observed in the Th group because previous studies indicate that increased NK cell activity may negatively affect longer-term outcomes in sepsis (28). However, our study did not examine NK cell counts after 48 h post-CLP and, therefore, cannot definitively assign a causal relationship with the later tramadol deaths. In these studies, we have only investigated one organ for activated NK cell numbers and have not yet evaluated the actual activity of the NK cells. Future studies are needed to fully define the effects of high-dose tramadol on NK cells.

Although our studies showed few significant differences relative to negative control, there were a number of differences between analgesic treatment regimens. The most consistent differences were seen between mice treated with buprenorphine and high-dose tramadol. The CLP mice treated with high-dose tramadol had significantly greater mortality than buprenorphine-treated mice, accompanied by higher plasma cytokine levels and significantly decreased WBC parameters at 48 h post-CLP. The low peripheral WBC counts, higher cytokine levels, and increased NK cell counts at 48 h may partially explain the reduced survival in the high-dose tramadol-treated mice as compared with buprenorphine. Because similar effects were not seen at the earliest time points, these differences may represent a cumulative dose effect. It should be noted that our study used fairly high doses of both tramadol and buprenorphine, as well as multiple administrations (up to four doses), to maximize the potential effects of the drugs. However, simple laparotomy may not require this level of analgesia in mice, and many IACUCs will approve similar studies with one or two doses of analgesics. In addition, our studies used isoflurane, a gas anesthetic that does not supply postoperative analgesia. Some rodent studies use the injectable combination of ketamine/xylazine. Xylazine has some analgesic effects and, therefore, lower doses of postoperative opioids could be used in combination with it. With lower doses and/or a reduced number of doses, differences in immune parameters and mortality rates may be even less likely to occur.

For most of our experiments, our treatment groups consisted of 12 mice, a sample size similar to that used in many sepsis studies. To evaluate immune parameters, the statistical analysis was designed to increase the power to detect differences across treatments at that sample size; however, this was done at the expense of not adjusting the type I error rate for multiple comparisons. Therefore, our study had the inherent risk of highlighting significant differences that may actually be caused by chance alone. Because of this consideration, actual P values are presented and more confidence that true differences exist between treatments should be given to comparisons with low P values (e.g., <0.01). For the mortality studies, the results indicated some nonsignificant variations from the negative controls for both high-dose tramadol- and buprenorphine-treated mice. Power analysis of these data revealed that the sample sizes needed to detect a significant difference (2-sided, α = 0.05) in mortality rates for negative-control versus high-dose tramadol- or buprenorphine-treated mice would be approximately 24 mice per group and 84 mice per group, respectively, or two and seven times the sample sizes used in this study. These results suggest that differences in mortality rates were minor but might be confounding if studies use large sample sizes.

Our study plan and statistical analysis were designed to maximize detection of differences in select immune parameters after treatment with specific opioids in a CLP model. Even with this design, the studies detected only a handful of significant differences between control mice and those treated with analgesics, and some of these may represent spurious associations because of the large number of statistical comparisons performed. None of the treatment regimens assessed in this study produced significant differences in immune parameters compared with negative controls that remained evident or consistent over time. Given the increased mortality and differences observed at 48 h compared with other treatments, as well as the significant differences in splenic NK cells, the authors do not recommend the use of the high dose of tramadol in sepsis studies. However, the results suggest that judicious use of one or two doses of buprenorphine or low-dose tramadol is unlikely to change outcomes significantly compared with no treatment in similarly designed studies with comparable sample sizes and immune parameters. More consistent differences were seen across the analgesic treatments, indicating that caution needs to be applied when comparing studies and that this information must be clearly presented in the Materials and Methods section of articles.

Of course, there are many other immune parameters that were not directly assessed in this study; importantly, the functionality of specific leukocytes or leukocyte subsets was not assessed. In addition, investigating the potential effects of analgesics across every permutation of the CLP model was beyond the scope of this study. However, it is possible that analgesics may lead to different results in CLP models that vary from the model presented here in any number of important parameters (e.g., needle size, puncture number, anesthetics, antibiotics, etc). In addition, females of an outbred stock of mice were studied in this investigation. It is known that female mice are more resistant to sepsis than male mice and mice of various strains can have highly variable responses in sepsis research (29), therefore extrapolation of results across sex and/or mouse strain should be done cautiously. The results reported here represent an initial investigation of the effects of opioid analgesics on a subset of immune parameters in a single CLP model of sepsis, and global recommendations cannot be made at this time. Given the increasing concerns for animal welfare, further research would be useful for IACUCs, investigators, and veterinarians to facilitate informed decisions regarding analgesic use in surgical models of sepsis.


1. Deitch EA: Rodent models of intra-abdominal infection. Shock 24(Suppl 1):19-23, 2005.
2. Buras JA, Holzmann B, Sitkovsky M: Animal models of sepsis: setting the stage. Nat Rev Drug Discov 4:854-865, 2005.
3. Hubbard WJ, Choudhry M, Schwacha MG, Kerby JD, Rue LW 3rd, Bland KI, Chaudry IH: Cecal ligation and puncture. Shock 24:52-57, 2005.
4. Nemzek JA, Hong-Yan X, Minard AE, Bolgos GL, Remick DG: Humane endpoints in shock research. Shock 21:17-25, 2004.
5. Public Health Service: Public Health Service Policy on Humane Care and Use of Laboratory Animals. Washington, DC: US Department of Health and Human Services, p 13, 2002.
6. Consales G, De Gaudia AR: Sepsis-associated encephalopathy. Minerva Anestesiol 71:39-52, 2005.
7. Molina PE: Opioids and opiates: analgesia with cardiovascular, haemodynamic and immune implications in critical illness. J Intern Med 259:138-154, 2006.
8. Vallejo R, de Leon-Casasola O, Benyamin R: Opioid therapy and immunosuppression: a review. Am J Ther 11:354-365, 2004.
9. Hilburger ME, Adler MW, Truant AL, Meissler JJ Jr, Satishchandran V, Rogers TJ, Eisenstein TK: Morphine induces sepsis in mice. J Infect Dis 176:183-188, 1997.
10. Angus DC, Linde-Zwirble WT, Lidicker J, Clermont G, Carcillo J, Pinsky MR: Epidemiology of severe sepsis in the United States: analysis of incidence, outcome, and associated costs of care. Crit Care Med 29:1303-1310, 2001.
11. Vender JS, Szokol J, Murphy GS, Nitsun M: Sedation, analgesia, and neuromuscular blockade in sepsis: an evidence-based review. Crit Care Med 32:554-561, 2004.
12. Toth LA: Defining the moribund condition as an experimental endpoint for animal research. ILAR J 41:72-79, 2000.
13. Interagency Research Animal Committee: US Government Principles for the Utilization and Care of Vertebrate Animals Used in Testing, Research, and Training. Federal Register, May 20, 1985. Washington, DC: Office of Science and Technology, 1985.
14. Sacerdote P: Opioids and the immune system. Palliat Med 20:9-15, 2006.
15. Remick DG, Bolgos GR, Siddiqui J, Shin J, Nemzek JA: Six at six: interleukin-6 measured at 6 h after the initiation of sepsis predicts mortality over 3 days. Shock 17:463-367, 2002.
16. Hawk CT, Leary S, Morris T: Formulary for Laboratory Animals, 3rd ed. Ames, IA: Wiley-Blackwell Inc, p 19, 2005.
17. Sacerdote P, Bianchi M, Manfredi B, Panerai AE: Effects of tramadol on immune responses and nociceptive thresholds in mice. Pain 72:325-330, 1997.
18. Shirzad H, Shahrani M, Rafieian-Kopaei M: Comparison of morphine and tramadol effects on phagocytic activity of mice peritoneal phagocytes in vivo. Int Immunopharmacol 9:968-970, 2009.
19. Barcia E, Martin A, Luz Azuara M, Sanchez Y, Negro S: Tramadol and hyoscine N-butyl bromide combined in infusion solutions: compatibility and stability. Support Care Cancer 15:57-62, 2007.
20. Nemzek JA, Siddiqui J, Remick DG: Development and optimization of cytokine ELISAs using commercial antibody pairs. J Immunol Methods 255:149-157, 2001.
21. Martucci C, Panerai AE, Sacerdote P: Chronic fentanyl or buprenorphine infusion in the mouse: similar analgesic profile but different effects on immune responses. Pain 110:385-392, 2004.
22. Carrigan KA, Saurer TB, Ijames SG, Lysle DT: Buprenorphine produces naltrexone reversible alterations of immune status. Int Immunopharmacol 4:419-428, 2004.
23. Piersma FE, Daemen MARC, Bogaard AE, Buurman WA: Interference of pain control employing opioids in in vivo immunological experiments. Lab Anim 33:328-333, 1999.
24. Franchi S, Panerai AE, Sacerdote P: Buprenorphine ameliorates the effect of surgery on hypothalamus-pituitary-adrenal axis, natural killer cell activity and metastatic colonization in rats in comparison with morphine or fentanyl treatment. Brain Behav Immun 21:767-774, 2007.
25. Wang G, Weng Y, Ishiguro Y, Sakamoto H, Morita S: The effect of tramadol on serum cytokine response in patients undergoing pulmonary lobectomy. J Clin Anesth 17:444-450, 2005.
26. Gaspani L, Bianchi M, Limiroli E, Panerai AE, Sacerdote P: The analgesic drug tramadol prevents the effect of surgery on natural killer cell activity and metastatic colonization in rats. J Neuroimmunol 129:18-24, 2002.
27. Topcu I, Ekici NZ, Isik R, Sakarya M: The effects of tramadol and fentanyl on gastrointestinal motility in septic rats. Anesth Analg 102:876-881, 2006.
28. Zeerleder S, Hack CE, Caliezi C, van Mierlo G, Eerenberg-Belmer A, Wolbin A, Wuillenmin WA: Activated cytotoxic T cells and NK cells in severe sepsis and septic shock and their role in multiple organ dysfunction. Clin Immunol 116:158-165, 2005.
29. Deitch EA: Animal models of sepsis and shock: a review and lessons learned. Shock 9:1-11, 1998.

Opioids; analgesia; sepsis models; animal welfare; cytokines

Supplemental Digital Content

©2010The Shock Society