Effects of Mild Perioperative Hypothermia on Cellular Immune Responses 

Sixty patients, classified as American Society of Anesthesiologists physical status I or II, were included in this study after we obtained approval from the hospital human studies committee and informed consent from the patients. Patients were hospitalized for abdominal surgery (for details, see Table 1) and were randomly assigned to two thermal care groups, routine care or forced-air warming, with 30 patients in each group. In the forced-air warming (normothermia) group, skin warming was initiated 45 min before anesthesia was induced, using an upper-body forced-air cover, which delivers air at 40 [degree sign]C (Bair Hugger, Augustine Medical, Eden Prairie, MN). Warming was continued throughout the operation and for 1 h after operation. In this group, intravenous fluids were warmed to 37 [degree sign]C. In the second (hypothermia) group, active skin and fluid warming were not used.

On the day of the surgery, patients were premedicated with 2 mg oral lorazepam 90 min before induction of anesthesia, followed by 2 or 3 mg midazolam given intravenously when they arrived in the operating ward. General anesthesia was induced in both groups of patients, as follows: 2 to 2.5 mg/kg propofol, 1.5 to 2 [micro sign]g/kg fentanyl, and 0.1 mg/kg vecuronium bromide, all given intravenously. After induction, anesthesia was maintained with nitrous oxide (up to 70%) and isoflurane, titrated to maintain mean arterial blood pressure within 20% of baseline values (end-tidal concentration, 0.6 - 1.2%). Additional vecuronium and fentanyl (up to 5 [micro sign]g/kg) were given as needed. Patients were hydrated during the surgery by crystalloid (15 ml [middle dot] kg^-1[middle dot] h^-1).

Perioperative core temperature was assessed by a temperature probe inserted into the esophagus, in accordance with the technique described by Kaufman. [9]During surgery, electric activity of the heart, noninvasive blood pressure, heart rate, pulse oximetry, and end-tidal carbon dioxide level were monitored. Only patients who did not require blood transfusion in the perioperative period were included in this study. Venous blood samples (20 ml) were collected when the premedication was given (baseline value), at the end of surgery, and 24 h and 48 h after the operation.

Within 1 h after blood was harvested, peripheral blood mononuclear cells (PBMC) were isolated from heparinized blood on a Ficoll-Paque gradient. The cells were suspended in heat-inactivated newborn calf serum (Biological Industries, Beith Haemek, Israel) containing 10% dimethyl sulphoxide (Sigma Chemical Co., St. Louis, MO) and frozen at -70 [degree sign]C until use. On the day of assay, within 1 month of cell freezing, cells were thawed quickly; washed three times in RPMI-1640 medium containing 1% penicillin, streptomycin, and nystatin; and supplemented with 10% heat-inactivated fetal calf serum referred to as complete medium, and their viability was tested by trypan blue dye exclusion. Cells were used only if the viability was > 95%. Several studies have indicated that freezing of PBMCs could alter the characteristics of these cells. [10]In a methodologic study, we compared cytokine production (IL-1 [small beta, Greek] and IL-6) of freshly isolated and frozen cells obtained from the same patients 90 min before and 24 h after surgery. Although there were differences in cytokine production between the freshly obtained and frozen cells, all comparisons between the two time points and between the two thermal care groups yielded similar results in the freshly isolated and cryopreserved cells. That is, whenever there were significant differences between the two time periods in the freshly obtained cells, significant differences in the same direction were also found in the frozen cells. Similarly, if no difference was found in freshly obtained cells, no difference was found in the frozen cells.

Peripheral blood mononuclear cells (2x10⁶) suspended in 1 ml complete medium were incubated in the absence or presence of 10 [micro sign]g/ml lipopolysaccharide (LPS)(Escherichia coli, LPS; Sigma Chemical Co.). At 24 h, cells were removed by centrifugation, and culture media were collected and maintained at -70 [degree sign]C until assay.

Peripheral blood mononuclear cells (2x10⁶) suspended in 1 ml complete medium were incubated in the presence of 0.2% phytohemagglutinin (PHA-M; Difco, Detroit, MI). At 48 h, cells were removed by centrifugation, and culture media were collected and maintained at -70 [degree sign]C until assay.

The concentrations of IL-1 [small beta, Greek] and IL-6 in the supernatants and in the sera samples were tested using ELISA kits specific for these human cytokines (Genzyme Corp., Cambridge, MA; Pharmingen, San Diego, CA), as described in the manufacturer guidelines. The detection levels of these cytokines in the assays used were 30 pg/ml for IL-1 [small beta, Greek] and 1 pg/ml for IL-6.

Interleukin-2 activity in the supernatants was determined by bioassay using the IL-2-dependent cell line CTLL 2 (purchased from the ATCC (#TIB-214), Rockville, MD), as described previously. [11]Briefly, 5x10 (⁴) CTLL 2 cells (for IL-2) were incubated for 48 h with supernatant samples (four serial dilutions, each in triplicates) in 96 flat-bottomed microtiter plates. Methyl-(³) H-thymidine (5 Ci/mmol; Rotem Industries LTD, Beer-Sheva, Israel) were added (0.5 [micro sign]Ci/well) 18 h before harvesting. CTLL 2 cells will not grow and will not incorporate methyl-(³) H-thymidine unless IL-2 is present in the culture media, and their growth is related to the level of IL-2 present in the samples. The results were calculated as units per milliliter using a computer program that compared the slopes obtained for the four dilutions of the sample with the standard curve (IL-2-containing preparations, composed of supernatants of rat splenocytes stimulated with concanavalin A [ConA]), which is arbitrarily considered 1,000 units.

Cytotoxicity was assessed by the standard chromium specific release assay, as described before. [12]Briefly, natural target cells for human natural killer (NK) cells (K562 erythroleukemia cell line) were used as target cells, and PBMC containing NK cells were used as effector cells. K562 cells were labeled by incubation for 1 h with⁵¹Cr, during which⁵¹Cr was taken up into cell cytoplasm. Labeled target cells were coincubated with PBMC in 96-well plates, for 4 h at 37 [degree sign]C, at a final effector-to-target ratio of 50:1. All reactions were done in triplicate. During the incubation, K562 cells were lysed by NK cells and released the⁵¹Cr into the supernatant. After 4 h of incubation, the plates were centrifuged, the supernatant samples were collected, and the radioactivity was assessed in a gamma counter (LKB, Turku, Finland). The percentage specific release was calculated using the following formula:(radioactivity level in the samples - spontaneous release)/(maximum radioactivity level of the labeled cells - spontaneous release)x100. Aliquots of 100 [micro sign]l supernatants from wells in which K562 cells were incubated in either media alone or in triton X-100 served to determine spontaneous and maximum release, respectively.

In this assay, PBMCs were incubated for 3 days with one of three mitogens, which activate and stimulate growth in certain lymphocyte populations, as follows: phytohemagglutinin stimulates primarily mature T cells, ConA stimulates both mature and immature T cells, and pokeweed mitogen (PWM) stimulates both T and B cells. Aliquots (0.1 ml) of PBMC suspension (2x10^6/ml) were added into each well of 96-well plates (flat bottom; Nunc, Roskilde, Denmark) containing 0.1 ml complete medium, PHA (PHA-M, 2%; Difco), ConA (10 [micro sign]g/ml), or PWM (20 [micro sign]g/ml; Sigma). Cultures set in triplicates were incubated for 3 days. For the last 18 h before harvesting, methyl-(³) H-thymidine (5 Ci/mmol) was added (0.5 [micro sign]Ci/well) to each well. Tritiated thymidine was taken up by stimulated and thus rapidly dividing cells. At the end of 72 h, plates were centrifuged and radioactivity was assessed in the cells using a LKB liquid scintillation counter (model 3380). The amount of radioactivity is proportional to the number of dividing cells.

Tumor necrosis factor content in the samples was determined by its cytotoxic effect on the fibrosarcoma cell line A-9. Cells were seeded in 96-well plates (Nunc) at 3x10^4/well, in 0.1 ml Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (Biological Industries). After 24 h of incubation at 37 [degree sign]C, supernatants were removed and replaced with 0.1-ml samples from the PBMC supernatants at various serial dilutions. After an additional 24 h of incubation, cell survival was assessed colorimetrically using the vital dye neutral red (Sigma Chemical Co.), which was taken up by living and viable cells only. Cell survival is reversely proportional to TNF level in samples of PBMC supernatants. TNF-[small alpha, Greek] cytotoxic activity in the samples was assessed by comparing the survival of cells incubated with the PBMC supernatants with the survival of cells incubated with a known amount of recombinant human TNF-[small alpha, Greek](Pepiotech, Rocky Hill, NJ). The detection level of this assay was 1 pg/ml.

Plasma cortisol levels were determined by radioimmunoassay (Diagnostic System Laboratories, Webster, TX).

The number of observations varies slightly among assays as a result of an occasional missing sample or occasional problem in a particular assay. Data were analyzed using analysis of variance for repeated measures. Post hoc comparisons were conducted as appropriate, correcting for multiple comparisons. [13]Probability values < 0.05 were considered significant. The results are expressed as mean +/− SD.

(Table 1) shows the characteristics of patients and surgery. No significant differences were observed between the two thermal care groups in age, weight, duration of surgery, and amount of anesthetics administered (Table 1).

Analysis of variance revealed significant main effects of thermal care (between groups; F sub (1,58)= 86.836), of time periods (within factor; F sub (9,522)= 246.798), and significant interaction (F sub (9,522)= 33.807). Mean core temperature was significantly less in the hypothermia group throughout the surgery, from 15 min after the beginning of surgery until 3 h after surgery (Figure 1). At 24 and 48 h after surgery, body temperature did not differ significantly between the two groups.

Analysis of variance tests revealed significant interaction of thermal care by time periods for ConA, PHA, and PWM (F sub (3,156)= 2.85, F sub (3,141)= 3.03, F sub (3,153)= 4.07, respectively). Multiple comparisons revealed no significant differences between the two thermal groups in presurgery levels of mitogens-induced proliferation. ConA-, PHA-, and PWM-induced proliferation was significantly suppressed in the hypothermia group, at 24 h after surgery, compared with the normothermia group. This suppression was still significant at 48 h for ConA, but not for PHA and PWM (Figure 2). In the normothermia group, mitogens-induced proliferation remained stable at all time periods.

A significant interaction of the thermal group by time periods was revealed for IL-1 [small beta, Greek] and IL-2 production (F sub (3,153)= 3.03, F sub (3,159)= 3.01, respectively). Production of both cytokines was significantly higher 24 h after surgery in the normothermia patients compared with the hypothermia patients, as revealed by multiple comparisons (Figure 3). No significant differences were observed between the two thermal care groups, at any time period, for IL-6 and TNF-[small alpha, Greek] production. There was a significant time period effect for both IL-6 and TNF-[small alpha, Greek] production (F sub (3,168)= 6.87, F sub (3,159)= 4.58, respectively). Production of both cytokines was significantly elevated at 48 h compared with baseline values, as revealed by contrast analyses (Figure 3).

There was a significant effect of time periods (F sub (3,150)= 15.98). Natural killer cell activity was significantly suppressed at 24 h after surgery in both thermal care groups, as revealed by contrast analyses. After 48 h, NK cell activity almost returned to control values in both groups of patients. No significant difference in NK cell cytotoxicity was found between the two groups at any time period (Figure 4).

There was a significant effect of time periods (F sub (3,129)= 99.76), and significant interaction of thermal care by time periods (F sub (3,129)= 6.01). Further analysis revealed that cortisol blood levels were significantly increased at the end of the surgery and at 24 h after operation. At 48 h after surgery, cortisol levels almost returned to baseline values. Plasma cortisol levels were significantly elevated in the hypothermia patients as compared with the normothermia group at 24 h after surgery (Figure 5).

The perioperative period is characterized by alterations of the immune system. [8]Many known factors may contribute to the observed alterations in the immune response during the perioperative period. These factors include type and duration of surgery, type of anesthesia, blood transfusion, and neuroendocrine changes. [8]The greater the extent of the surgical trauma, the greater the immune alterations. It has been suggested that anesthesia contributes to perioperative immune alterations. We recently reported that large-dose fentanyl anesthesia prolongs the NK suppression observed in the perioperative period, compared with small-dose fentanyl. [14] 

Hypothermia could induce immune alterations in humans and in animals. This has been shown in studies of hypothermia not related to surgery. Various components of the immune system are affected by the exposure to cold, such as decreased cell-mediated immunity [15]and NK cell activity [16]; reduced number of thymocytes, splenocytes, and CD4+ cells; suppression of B lymphocytes; and defective function of T lymphocytes. [16-18]Hypothermia has also been associated with suppressed phagocytic activity, including decreased migration of polymorphonuclear cells, reduced superoxide anion production, and reduced bacterial killing. [19,20] 

It is possible that hypothermia contributes to the immune alterations observed perioperatively. The effect of hypothermia on immune function in this period has not been studied extensively. Perioperative hypothermia increased the rate of wound infections and prolonged hospitalization. [3]These consequences might be related to decreased initial local responses induced by mild perioperative hypothermia, including thermoregulatory vasoconstriction, decreased oxygen delivery to wounded tissue, [21]inhibition of oxidative killing by neutrophils, [7]and reduced collagen deposition. [3]The present findings indicate that hypothermia may contribute to systemic suppression of immune reactivity in the perioperative period.

In the current study, we tried to determine whether maintaining intraoperative normothermia attenuates the immune changes associated with surgery. Perioperative warming of patients, by using warm forced-air and fluids, maintained normal core temperature, whereas in the unwarmed patients core temperature decreased approximately by 1 [degree sign]C during the operation. Differences in immune responses were observed between the two study groups.

Mitogen stimulation of lymphocytes in vitro is believed to mimic the series of events that occurs in vivo after their stimulation by specific antigens. In the present study, there was a significant suppression of mitogenic responses (PHA, ConA, and PWM) in the hypothermia group 24 h after surgery, which was still evident at 48 h in response to Con A. The mitogens PHA and ConA activate T cells, whereas PWM stimulates both T and B cells, thus indicating that the suppressive effects of hypothermia involve various lymphocyte subpopulations. Mitogenic responses remained unchanged throughout the observation period in the normothermia group.

Twenty-four hours after surgery there was a significant decrease in IL-2 production in the hypothermia group compared with the normothermia group. This finding is in accordance with the results of previous reports in which diminished IL-2 secretion was found in the postoperative period. [22]However, in patients in the normothermia group, suppression of IL-2 production was prevented.

Interleukin-2 is a growth factor for T lymphocytes and consequently stimulates production of T-cell-derived cytokines, including interferon-gamma, and TNF, as well as several B-cell-activating cytokines, such as IL-4, IL-5, and IL-6. Because IL-2 plays a central role in various immune responses, reduced production of this cytokine in patients of the hypothermia group may impair their immune defense mechanisms and increase their susceptibility to infections. Surgical stress during cancer resection causes a discharge of tumor emboli into the systemic circulation and enhances the growth of existing micrometastases. [23]Interleukin-2 has been shown to stimulate NK-lymphokine-activated killer cells and tumor-infiltrating lymphocytes, [24,25]so IL-2 may play a role in the defense mechanisms against cancer dissemination. That IL-2 production was significantly suppressed in hypothermic patients 24 h after operation may prove to be important to the spread of metastases.

Interleukin-1 [small beta, Greek] and -6 and TNF-[small alpha, Greek] are known as “proinflammatory cytokines,” because they exert a variety of effects to accelerate inflammation. In the current study, an increased production of all three cytokines was observed postoperatively. This increased production reached maximal values at 24 h after surgery for IL-1 [small beta, Greek] and at 48 h after the surgery for IL-6 and TNF-[small alpha, Greek]. Significant differences between the two thermal groups were observed only for IL-1 [small beta, Greek] production, which was significantly higher in the normothermia group, compared with the hypothermia group, 24 h after surgery. The observed increase in proinflammatory cytokine production after surgery confirm the findings of earlier studies.

The proinflammatory cytokines, including IL-1, IL-6, and TNF-[small alpha, Greek], have been shown to promote wound healing. [26]Interleukin-1 [small beta, Greek] has been shown to accelerate wound healing, presumably because of its ability to induce angiogenesis and fibroblast activation. [27]Together these findings suggest that delayed wound healing after perioperative hypothermia [3]could be accounted for, at least in part, by reduced IL-1 [small beta, Greek] production in hypothermic patients.

Tumor necrosis factor and IL-1 can elicit production of tissue factor and in this way shift the hemostatic balance toward coagulation and away from anticoagulation. [28]Thus cytokines may reduce intraoperative blood loss and blood requirements in this period.

Natural killer cell cytotoxicity was suppressed in both values at 48 h. There was no significant difference between the two groups. This suppression is in accordance with our earlier findings. [14] 

In conclusion, the present findings indicate that intraoperative hypothermia can induce suppression of immune function and thus may exacerbate the complications of surgery. Maintaining intraoperative normothermia can reduce immune alterations perioperatively and may improve patients' outcome.