Morphine Inhibits Spontaneous and Cytokine-enhanced Natural Killer Cell Cytotoxicity in Volunteers

This study was approved by the Dartmouth-Hitchcock Medical Center Committee for Protection of Human Subjects and Institutional Animal Care and Use Committee.

Participants were paid, healthy volunteers, aged 24-45 yr, of either sex; were receiving no long-term medications; and had no history of opioid intolerance, substance abuse, or immunologic deficiency.

Participants served as their own control for each immunologic assay and subsequent analysis. Preliminary data in six subjects who received oral morphine for 36-60 h [6]showed an average (plus/minus SD) decrease in peripheral blood mononuclear cell (PBMNC) ADCC of 9.5% (plus/minus 6.6) specific cytotoxicity. Although NKCC, not ADCC, was the primary outcome variable in this study, we calculated that, with ten subjects in the high-dose study group and assuming that NKCC would be comparably depressed, we would have 95% power to detect a significant decrease in NKCC with a two-sided 5% level test.

The study was conducted in two sequential phases. Peripheral blood NKCC was the primary study variable in both phases. During the first phase, participants received intravenous morphine at a dose of 0.025 mg/kg loading dose followed by 0.015 mg *symbol* kg sup -1 *symbol* h sup -1 infusion ("low" dose) for 24 h. This phase of the study tested reproducibility of earlier experimental findings regarding morphine effects on ADCC and leukocyte antibody Fc receptor expression [6]and provided experience with the study protocol before initiating a higher-dose morphine study. Intergroup comparisons of results were not planned because of the nonblinded, sequential nature of the study and insufficient power to detect significant intergroup differences. During the second phase of the study, participants received morphine at a dose of 0.05 mg/kg loading dose followed by 0.03 mg *symbol* kg sup -1 *symbol* h sup -1 ("high" dose) designed to approximate more closely the analgesic doses administered to patients in pain. During the second phase, lymphocytes from participants were also tested for lymphocyte infectivity with HIV. The study protocol included one 30-mg oral dose of morphine (MS Contin, Purdue Frederick, Norwalk, CT) taken the night before intravenous morphine to test subjects for sensitivity to opioid side effects before initiation of intravenous morphine. On the following day, participants were admitted to the hospital, and a peripheral intravenous infusion of lactated Ringer's solution was initiated at 50 ml/h. As soon as intravenous access was established, the morphine infusion was started at the doses described using a commercially available infusion device (PCA-plus, Abbott Lifecare) with a morphine concentration of 1 mg/ml. The morphine infusion was initiated between 9:00 and 10:00 AM. The study protocol allowed a one-time, 50% decrease in the morphine infusion rate as a treatment for morphine side effects. Continued side effects were treated by discontinuation of the morphine infusion, pharmacologic therapy if necessary, and withdrawal from the study. After initiation of the morphine infusion, participants remained in hospital for 24 h. At the end of 24 h the infusion was stopped and participants were discharged from the hospital. Peripheral blood for analysis was drawn during a baseline period the week before morphine exposure (baseline), 2 h after the initiation of intravenous morphine (2 h), at the end of the morphine infusion (24 h), 24 h after termination of the morphine infusion (48 h), and 7-10 days after termination of the morphine infusion (8 day).

PBMNCs were isolated as described [6]and resuspended in RPMI-1640 supplemented with 10% fetal calf serum ("serum-free" medium to designate that it did not contain human autologous serum) at the cell concentration required for subsequent assays. After isolation, PBMNCs used for NKCC assays were either suspended immediately in serum-free medium, or in medium containing 50% autologous serum, and used in the NKCC assay, or incubated overnight in serum-free medium at 37 degrees Celsius and 5% CO sub 2 with or without gamma-interferon (IFN-g) (Genentech, South San Francisco, CA) at a concentration of 10 U/ml. Serial dilutions achieved the final desired effector:target (E:T) ratios. Similar isolation and testing procedures were used for the ADCC assays.

The human myeloid cell line, K-562, was used as the target cell in the NKCC assay. Aliquots of K-562 cells containing 10⁶cells were labeled with⁵¹Chromium, washed, and resuspended to a final concentration of 10⁵cells/ml. Aliquots of 100 micro liter of the resulting suspension (10⁴cells) were pipetted into 96-well round-bottom plates containing effector cells at the desired E:T ratio. After incubation at 37 degrees Celsius with 5% CO²for 4 h, 100 micro liter supernatant was removed from each well and counted for 1 min on a gamma counter to quantify⁵¹Chromium release. The percentage of isotope released was used as a measure of cytotoxicity and was calculated as Equation 1where CPM (experimental) = counts released after incubation of effector cells with target cells; CPM (maximum) = counts released by lysis of 1 x 10⁴target cells with detergent; and CPM (control) = counts released after incubation of target cells in medium alone. Results are expressed as percent specific lysis at a single E:T ratio and as lytic units (20% specific lysis) per 10⁷effector cells. [10]Specific cytotoxicity results represent the mean of three determinations.

Chicken erythrocytes (CE) were used as target cells for the ADCC assay and were obtained fresh on the day of the experiment by venipuncture, washed, labeled with⁵¹Chromium in a mixture of 15 micro liter packed CE and 50 micro liter fetal calf serum. Labeled target cells were resuspended to a concentration of 0.5 x 10⁶cells/ml. Washed packed ox cells were also added to the target cells to increase cell density and keep spontaneous lysis low. Rabbit anti-CE antibodies were diluted with medium to final antibody concentrations of 2, 1, 0.1, and 0.01 micro gram/ml. Fifty micro liter each of medium containing CE, effector cells, and rabbit anti-CE antibodies were added to 96-well round-bottom plates in triplicate at an E:T ratio of 20:1. After incubation in 5% CO²at 37 degrees Celsius, 75 micro liter of the supernatant was harvested into disposable culture tubes and counted for 1 min in a gamma counter. The percentage of ADCC was determined using a calculation similar to that used for NKCC. Control conditions for antibody independent and antibody-only killing had 0% killing.

Quantitation of Fc receptors was performed by an indirect immunofluorescent procedure using the following monoclonal antibodies at saturating concentrations: 22 (Medarex, Annandale, NJ), which binds specifically to Fc receptor I; IV.3 (Medarex, Annandale, NJ), which binds specifically to Fc receptor II; and 3G8 (obtained from Dr. Jay Unkeless, Mt. Sinai Hospital School of Medicine, New York, NY), which binds specifically to Fc receptor III. The immunoglobulin G1 myeloma, P3 (a gift from Dr. Michael Fanger, Dartmouth Medical School), was used as an isotype control for nonspecific binding. Mean fluorescence intensity was quantified by flow-cytometric analysis using a fluorescence-activated cell sorter (Becton Dickenson, San Jose, CA). A standard curve was constructed with mean fluorescence intensities of six fluorescein-labeled latex beads and their respective fluorescein concentrations (Flow Cytometry Standards, Research Triangle Park, NC). The mean fluorescence intensity of the effector cells was translated to the number of antibodies bound per cell (fluorescein isothiocyanate-conjugated goat anti-mouse Fab'²per cell) from a standard curve. The number of antibody molecules bound per cell in each of the control conditions (autofluorescence and second antibody only) was always less than the number of antibody molecules bound per cell with the isotype control. Antibody molecules bound per cell were corrected for nonspecific binding by subtracting the level of P3 binding from the binding of monoclonal antibodies 22, IV.3, and 3G8.

For analysis of morphine effects on HIV infectivity, PBMNC were isolated and stimulated in culture for 3 days in the presence of phytohemagglutinin, washed extensively, and infected with the HIV-1JR-FL strain. Cultures were maintained without phytohemagglutinin, but with exogenous interleukin-2 at a concentration of 1,000 U/ml for 1 week. Cells were fed midway through the second culture period (day 4) with fresh medium and interleukin-2. Viral production in each culture was assayed on day 7 by measuring p24 antigen concentration by enzyme-linked immunosorbent assay (HIVag assay, Abbott Laboratories, North Chicago, IL) and quantified using a standard quantitation panel (Abbott Laboratories).

Serum samples were frozen at -70 degrees Celsius and batched for analysis of morphine concentration in peripheral blood. The sample preparation method used a mixed-bed solid-phase extraction column. The eluate was dried in a Speed-Vac vacuum centrifuge (Savant, Farmingdale, NY) with heat. Chromatographic separations were performed using a cyano column maintained at 35 degrees Celsius in a water bath. A guard column packed with a cyano-bonded stationary phase placed before the analytical column and renewed frequently was used to maintain column efficiency. The high-performance liquid chromatography buffer was 50 mM sodium phosphate in 50% methanol, brought to pH 7.0 with phosphoric acid, and filtered before use. The flow rate was 0.7 ml/min. High-performance liquid chromatography-grade water and methanol were used. The detector used two porous graphite cells set at +0.4 and +0.8V and a guard cell at +0.9 V. Hydromorphone, added before the extraction step, was used as an internal standard for morphine determinations. The assay has a sensitivity of 0.5 ng/ml for both morphine and hydromorphone.

We computed mean relative differences (as percentages) from baseline and standard errors for the NKCC high-dose- and low-dose-treated assays separately at each of the four periods (2 h, 24 h, 48 h, and 8 day). These calculations were also done for percent specific cytotoxicity at specific E:T ratios. Repeated-measures analysis [11-13]was used to test whether the relative cytotoxicity at each time period was different from that at baseline. For the ADCC, Fc receptor, HIV infectivity, IFN-g stimulated NKCC, and NKCC serum assays, we computed the relative differences from baseline defined as the absolute difference divided by the baseline value reported as a percentage. We computed the relative difference because the absolute differences appeared to be proportional to the baseline cytotoxicity. Mean relative differences and standard errors were calculated for ADCC, Fc receptors, and HIV assays, and for IFN-g-stimulated NKCC and the NKCC autologous serum assays. We used a paired t test to evaluate differences in these measures from baseline for each group separately.

Twenty-three participants were entered into the study protocol (13 men and 10 women). Five (4 men and 1 woman) were withdrawn before completion of the morphine infusion because of morphine-related side effects, including anorexia, nausea, vomiting, and pruritus. Two participants were given a saline infusion in a blinded manner to evaluate the effect of hospitalization on NKCC. Sixteen participants completed the study protocol with morphine; 7 received the low dose and 9 received the high dose of morphine. Five participants required a decrease in the morphine infusion rate for treatment of nausea and vomiting, 2 in the low-dose study phase and 3 in the high-dose phase. Participants in the low-dose group received an average total of 0.35 mg/kg intravenous morphine (range 0.28 mg/kg-0.40 mg/kg). Participants in the high-dose group received an average total of 0.59 mg/kg intravenous morphine (range 0.44 mg/kg-0.77 mg/kg) resulting in a mean serum morphine concentration of 18.5 ng/ml (plus/minus 4.3 SE) at the end of the high-dose infusion. Except as noted above, there were no complications from the morphine infusion.

Morphine resulted in significant depression of PBMNC NKCC in both the low-dose and high-dose groups (Table 1). Depression of NKCC was observed by 2 h and was maximal at 24 h. Recovery of NKCC was apparent by the 48-h measurement period in the low-dose study group but remained significantly depressed in the high-dose study group. Depression of NKCC was significant at 24 h regardless of whether cytotoxicity was calculated as lytic units or expressed as percent specific lysis (Table 1and Figure 1). Suppression of unstimulated NKCC was observed both when effector cells were incubated in medium free of autologous serum (Table 1) and when effector cells were incubated in autologous, morphine-exposed serum separated at 24 h (Figure 2). NK cell suppression did not appear to be any greater in the presence of autologous serum than in serum-free medium. Suppression of NK activity in the presence of medium with serum was significant in both the low-dose and the high-dose study phases at the 24-h measurement period compared with baseline values (P < 0.05 for both doses). Effector cells incubated overnight with the stimulatory cytokine, IFN-g, demonstrated significantly increased cytotoxicity compared with unstimulated baseline NKCC (P < 0.001). However, morphine significantly decreased IFN-g-induced enhancement of NKCC in both the low-dose study phase and the high-dose study phase (Table 2). Finally, an infusion of saline given in a blinded manner had no effect on NKCC in the two subjects in which it was tested. In these two participants, baseline lytic units were 177 and 132, and 24-h lytic units were 189 and 191, respectively.

Morphine exposure in the low-dose phase of the study resulted in significant depression of ADCC similar to that previously reported. [6]Suppression of ADCC was significant at an intermediate antibody concentration of 0.1 micro gram/ml (Table 2) and was not observed at very low antibody concentrations, at which very little cytotoxicity was measured, or at very high antibody concentrations, at which high levels of cytotoxicity were always observed. gamma-Interferon-induced enhancement of ADCC was also slightly, but significantly, decreased (Table 2). There was no significant effect of morphine on Fc receptor expression on monocytes, lymphocytes, or polymorphonuclear cells (Table 3). The effect of morphine exposure on HIV infectivity was measured in lymphocytes drawn from six participants in the high-dose phase of the study (Table 4). The difference between the mean value of 42.8 ng/ml p24 measured with baseline cells versus 100.4 ng/ml p24 measured using cells obtained at 24 h was not statistically significant (P = 0.17).

Opioid-immune interactions are potentially important, given the widespread use of opioids, the central role of the immune system in a variety of diseases, and the introduction of new immunologic treatments for diseases that range from infection to cancer. [14-16]Results reported from this study document depression of peripheral blood spontaneous NKCC after in vivo morphine administration. This phenomenon may be dose-dependent because an earlier study reported no depression of NKCC after morphine doses that resulted in lower serum morphine concentrations. [6]In addition, although the current study did not attempt intergroup comparisons, NKCC remained significantly depressed for 48 h after high-dose morphine but had returned to normal by 48 h after low-dose morphine.

Depressed NK activity was partially restored by IFN-g treatment. Aside from anti-viral effects, [17]IFN-g has can activate monocyte and macrophages against a variety of pathogens [18]and also activates NK cells against tumor targets. [19]The latter effects have formed the basis of clinical trials of IFN-g as a treatment for infection [14]and advanced cancer. [19,20]Although our results suggest that in vivo morphine exposure interferes with activation of NK cells by IFN-g, the IFN-g exposure (and testing of NK function) was performed ex vivo and may not reflect in vivo effects of IFN-g. Resistance of splenic NK cells to IFN-g stimulation was recently reported in mice exposed to inhalational anesthetics before IFN-g stimulation of splenic NK cells. [21].

Morphine effects on NK activity were observed in the absence of autologous serum and after incubation of effector cells for as long as 18 h. This observation, and the observation that morphine also decreased ADCC-mediated lysis after low-dose exposure of participants, suggests a durable effect of morphine on peripheral blood effector cells.

Most of the information available on opioid-immune interactions comes from animal studies or from in vitro testing. This study was conducted in healthy volunteers for two reasons. First, diseases commonly associated with opioid use can alter immune studies. Immune studies performed in patients who require analgesia are difficult to interpret because of the coexisting condition that requires pain therapy. Major surgery, for example, is associated with immune suppression [22-24]as are many types of chronic cancer [25-28]and substance abuse. [29,30]Second, in vivo morphine exposure more closely approximates the physiologic conditions under which opioids may affect human immunity. Results from in vitro investigations fail to approach the clinical situation in which effector cells are exposed to opioids in vivo. In addition, morphine has active metabolites [31]that may alter immune function in vivo but escape detection with in vitro testing of the parent drug.

There are several potential mechanisms by which opioids may alter human immunity. Opioids may have an indirect effect on immune function through alterations in effector cell populations. A decrease in effector cell density in any immune compartment (thymus, spleen, peripheral blood, lymph node) will alter functional assays that use mixed cell populations taken from any one compartment and tested without subset analysis of different cell types. Animal studies document sustained narcotic effects on effector cell populations. For example, morphine pellet implantation in mice results in significant thymic and splenic atrophy with an increased ratio of CD4 sup +/CD8 sup + cells but a decreased total number of both cell populations. [32]This effect was observed within 24 h after pellet implantation. Similarly, long-term daily morphine administration to rhesus monkeys results in decreased CD16 sup + cells (NK cells) in peripheral blood along with suppression of CD4 sup + cells. [33]Alterations in lymphocyte subsets have been noted in HIV-negative heroin abusers. [30]Second, opioids can also act to directly suppress some components of the immune response such as the antibody response of splenocytes, [34]rosette formation by lymphocytes, [35]and respiratory burst activity of PBMNCs. [36]Third, opioids can suppress immunity through effects on the central nervous system. For example, intraventricular administration of very small amounts of morphine (20-40 micro gram) results in depressed NKCC similar to that observed after peripheral, subcutaneous administration of much larger morphine doses (30-50 mg/kg). [37]This suggests a central nervous system-mediated effect after peripheral injection. Peripheral administration of N-methylmorphine, which does not cross the blood-brain barrier, has no immunosuppressive effect, whereas microinjections of N-methylmorphine into the third ventricle of the brain mediate immune suppression. [38]Opioids also cause central nervous system-mediated release of corticosteroids, which may lead to immune suppression. [39]Fourth, opioids may interact with suppressor cell populations or endogenous immune mediators. Stimulated PBMNCs have reduced output of IFN-g when incubated with morphine at pharmacologic concentrations. This effect requires monocytes because monocyte-depleted populations did not demonstrate the effect. [36]Similarly, incubation of PBMNCs with pharmacologic concentrations of morphine increased release of transforming growth factor beta, an immune suppressive mediator. [40]Finally, opioids may affect immune function by suppression of endogenous, tonic enhancement of effector cell activity. For example, endogenous opioids such as beta-endorphin circulate in blood at measurable concentrations that have been shown to stimulate immune effector cells in vitro. [41,42]However, coincubation of PBMNCs with pharmacologic concentrations of morphine or the synthetic opioid, fentanyl, inhibits beta-endorphin-induced increases in NKCC in vitro. [6]In vivo observations correlate with these findings. [43].

Peterson et al. reported that PBMNCs taken from normal humans and cocultured with morphine in the presence of HIV-infected lymphocytes showed an increase in HIV infectivity. [5]This effect was observed with morphine concentrations at or near the pharmacologic concentration required for clinical analgesia. Enhanced HIV infectivity was reversed by opioid receptor antagonists and was of approximately the same magnitude as that observed after exposure of PBMNCs to interleukin-2, which is also known to increase HIV proliferation. [44]Enhanced HIV expression in an infected promonocytic cell line has also been observed after coculture of infected cells with morphine-exposed fetal brain cells. [45]These studies raise the question of whether in vivo morphine exposure could alter HIV infectivity of target cells and thereby either alter susceptibility to HIV infection or alter the course of the disease. Based on these reports, we analyzed lymphocyte HIV infectivity in six morphine-exposed participants. Overall, we were not able to confirm a significant change in lymphocyte HIV infectivity after administration of morphine to study participants.

In summary, healthy volunteers who received intravenous morphine had measurable depression of peripheral blood spontaneous and IFN-g-stimulated NKCC. These effects were observed after effector cell incubations in the absence of autologous serum separated at the time of the assay, which suggests a durable effect on effector cell numbers or function. The clinical implications of this and other opioid-immune interactions are unknown. No adverse immunologic consequences were observed in any of the subjects who participated in this study. The question of whether or not opioid-induced immune alterations have clinical effects in patients with specific diseases will require further study.

The assistance of Duane Nash, Esther Colby, Kim Coleman, and Jennifer Fullerton is gratefully acknowledged.