Opioids Inhibit Febrile Responses in Humans, Whereas Epidural Analgesia Does Not: An Explanation for Hyperthermia during Epidural Analgesia

With approval from the University of California, San Francisco, Committee of Human Research and informed consent, we studied eight healthy male volunteers, each on four study days. None was obese, was taking medication, or had a history of thyroid disease or Raynaud syndrome. Morphometric and demographic characteristics included: age, 27 ± 6 yr; height, 180 ± 8 cm; weight, 74 ± 8 kg; and body fat, 17 ± 5%.

The volunteers fasted for 8 h before arriving at the laboratory. They were minimally clothed and reclined on a standard operation table during the study. Ambient temperature was maintained near 22°C and relative humidity near 45%. To avoid circadian fluctuations, studies were scheduled at the same time each day.

An 18-gauge catheter was inserted in a left forearm vein for fluid and drug administration. Lactated Ringer’s solution at ambient temperature was infused at a rate of 100 ml/h. A 16-gauge catheter was inserted in the right median-cephalic vein for blood sampling.

On each study day, the volunteers were randomly assigned to either: (1) a control day when no opioid or epidural analgesia was given; (2) epidural analgesia using 0.2% ropivacaine alone; (3) epidural analgesia using 0.2% ropivacaine in combination with 2.0 μg/ml fentanyl; or (4) intravenous fentanyl at a target plasma concentration of 2.5 ng/ml. Volunteers were then given an intravenous injection of 50 IU/g of human recombinant IL-2 (elapsed time, 0), observed 2 h later by 100 IU/g of the drug (Chiron, Inc., Berkeley, CA). At least 2 weeks were allowed between the fentanyl day and the combined epidural ropivacaine–fentanyl day to minimize tolerance.

On the epidural days, a catheter was inserted into the epidural space via  the L2–L3 interspace. The epidural catheter was then injected with 2 ml lidocaine, 2%, with epinephrine 1:100,000. This test dose was followed in 5 min by 10–12 ml of 0.2% ropivacaine (Ropivacaine HCl; Astra, Inc., Westborough, MA) or ropivacaine in combination with 2 μg/ml fentanyl without epinephrine. The initial anesthetic dose was based on the volunteers’ heights and calculated to produce a dermatomal level of sensory blockade near T8–S1 bilaterally as determined by loss of cutaneous cold sensation and response to pinprick. Epidural analgesia was then maintained with the continuously administered drug at a rate of 8–12 ml/h to maintain the target sensory block level.

On the intravenous fentanyl day, fentanyl was administered using a pump (Ohmeda 9000; Ohmeda, Steeton, UK) programmed to target fentanyl blood concentrations of 2.5 ng/ml using a modification of the Kruger–Thiemer method 11and published data. 12This is a low concentration that produces only mild sedation and is similar to concentrations that may be used for labor or postoperative pain. 13,14To maintain an appropriate end-tidal carbon dioxide tension (Pco²), the volunteers were reminded to breathe during fentanyl administration, and, if necessary, oxygen was given via  nasal cannula to maintain oxygen saturation measured by pulse oximetry (Spo²) more than 95%. Fentanyl was infused for 8 h and then discontinued.

Core temperature was measured at the tympanic membrane using Mon-a-Therm thermocouples (Mallinckrodt Anesthesiology Product, Inc., St. Louis, MO). Mean skin-surface temperature was calculated from measurements at 15 area-weighted sites. 15Temperatures were recorded at every 5 min from thermocouples connected to Iso-Thermex thermometers having an accuracy of 0.1°C and a precision of 0.01°C (Columbus Instruments Corp., Columbus, OH).

Pupillary responses were measured to evaluate the pharmacodynamic effect of fentanyl. An infrared pupillometer (Fairville Medical Optics, Buckinghamshire, UK) was programmed to provide a 0.5-s, 130 candela/m²pulse of green light and to scan the pupil at the rate of 10 Hz for 2 s from the beginning of the light stimulus. We have previously described this method 16and used it to quantify opioid effect. 17The maximum reduction in pupil size during the 2-s scan identified the reflex amplitude. Ambient light was maintained near 150 lux, and the contralateral eye was covered during measurements.

Peripheral venous blood for fentanyl analysis was sampled just before IL-2 administration and at 1, 3, 5, and 7 elapsed h. The plasma samples were stored at −20°C until analysis by gas chromatography, using techniques described by Bjorkman and Stanski 18and Selinger et al . 19The limit of detection is near 0.2 ng/ml.

We evaluated IL-6, IL-8, and TNF α, as well as the antiinflammatory cytokine IL-10 hourly. 20–23Plasma IL-6 and IL-8 concentrations were measured by an enzyme-linked immunosorbent assay (Human interleukin-6 and interleukin-8 ELISA Kits; Toray Industries, Inc., Tokyo, Japan). TNF α concentrations were determined by a human immunoassay (Quantikine HS; R&D Systems, Minneapolis, MN). Plasma IL-10 concentrations were determined by a solid-phase enzyme-amplified immunoassay (IL-10 EASIA Kit; Medgenix Diagnostics S.A., Fleurus, Belgium). In each case, the assays were performed per  the manufacturers’ directions, and appropriate calibration curves were constructed. All are highly specific and sensitive over the range of observed values.

Heart rates and arterial oxygen saturation were monitored continuously using pulse oximetry (Biox 3700; Ohmeda, Salt Lake City, UT). Blood pressure at the ankle was determined oscillometrically (Critikon, Tampa, FL). End-tidal carbon dioxide concentrations and respiratory rates were monitored using a Capnomac Ultima (Datex Medical Instruments, Tewksbury, MA). All temperatures and hemodynamic data were recorded at 5-min intervals.

Febrile responses on each study day are presented as time-dependent changes. Specifically, we considered integrated core temperature, peak temperature, and the time-to-peak temperature. Values were integrated during 3–8 elapsed h, with respect to the mean temperature during the first elapsed hour. That is, we calculated the area under the temperature curve. This quantified the extent to which core temperature exceeded initial values (in °C · h) and is a standard way of expressing fever magnitude.

Ambient temperature and humidity, hemodynamic responses, end-tidal Pco², respiratory rate, Spo², and administered fluid volume on each study day were first averaged within each volunteer during 3–8 elapsed h; the resulting values were then averaged among volunteers.

Most results were compared using repeated-measures analysis of variance and Scheffé F test. Time-dependent changes were similarly evaluated. A chi-square analysis was used to evaluate the fraction of the temperature measurements in each group exceeding 38.0, 38.5, and 39.0°C. Results are presented as mean ± SD;P < 0.05 was considered statistically significant.

The volunteers were only mildly sedated by intravenous fentanyl. Total plasma fentanyl concentrations during administration of fentanyl averaged 1.5 ± 0.2 ng/ml on the intravenous fentanyl day, but only 0.2 ± 0.1 ng/ml when fentanyl was given epidurally. Pupil size and reflex amplitude were significantly reduced by intravenous fentanyl (table 1). There were no significant differences in ambient temperature, relative humidity, heart rate, blood pressure, end-tidal Pco², and Spo²on the four study days.

Core temperatures peaked at 38.7 ± 0.6°C on the control day, at 38.7 ± 0.7°C during epidural ropivacaine, at 38.7 ± 0.9°C with combined epidural ropivacaine–fentanyl, but only at 38.1 ± 0.7°C during intravenous fentanyl (table 2). Fentanyl significantly reduced the febrile response to pyrogen, decreasing integrated core temperature from 7.0 ± 3.2°C · h on the control day to 3.8 ± 3.0°C · h on the intravenous fentanyl day. In contrast, epidural ropivacaine and epidural ropivacaine–fentanyl did not inhibit manifestation of fever (fig. 1).

The fraction of core-temperature measurements that exceeded 38°C was halved by intravenous fentanyl. The fraction of measurements exceeding 38.5°C was reduced more than fivefold by intravenous fentanyl (table 3). Both these reductions were statistically significant (P < 0.001).

Plasma concentration of TNF α, IL-6, and IL-8 increased later, reaching peak concentrations after 4 or 5 elapsed h. All subsequently decreased to near-baseline values by 8 or 9 elapsed h. In contrast, IL-10 continued to increase throughout the study period. None of the cytokine concentrations differed significantly on the four study days.

Epidural analgesia is associated with a high incidence of hyperthermia, 1,5,8which provokes expensive and invasive interventions. 6,7Because passive hyperthermia and excessive heat production are unlikely causes, elevated body temperature in laboring and postoperative patients is presumably true fever resulting from infection, tissue damage atelectasis, and so forth. The conventional understanding is that hyperthermia associated with epidural analgesia in the same clinical situation is caused by epidural analgesia. However, this theory is inconsistent with our observation that epidural analgesia failed to influence febrile responses. Instead, our data suggest that fever is inhibited by opioid administration in the “control” subjects.

Although maximum and average core temperatures differed by less than 1°C when the volunteers were given fentanyl and epidural analgesia, integrated temperature was halved by low-dose fentanyl. More importantly, the fraction of core temperatures exceeding various temperatures was reduced by a factor of two to five. This is a critical outcome because temperatures exceeding specific values often trigger laboratory investigations for infection and even antibiotic administration. That intravenous fentanyl reduces the fraction of temperatures exceeding various threshold temperatures thus explains the high incidence of “fever work-ups” in laboring mothers and their children after epidural analgesia. 6,7It is also consistent with the fact that there is no evidence whatsoever suggesting that these additional infection investigations were otherwise justified or in any way influenced outcome. 2 

Published reports demonstrate a correlation between hyperthermia during epidural analgesia and clinical signs of infection. 24There is also evidence that hyperthermia during epidural analgesia is highly associated with placental inflammation. 25These authors concluded:“Epidural analgesia is associated with intrapartum fever, but only in the presence of placental inflammation. This suggests that the fever reported with epidural analgesia results from immune responses rather than the analgesia itself.” These results are consistent with our theory that hyperthermia during epidural analgesia is associated with inflammation (although not necessarily infection). Our results thus suggest that hyperthermia during epidural analgesia should be taken seriously and not considered a benign complication of the anesthetic technique per se . In contrast, threshold temperatures triggering investigations for infection should probably be reduced by approximately 0.5°C in patients given opioid analgesia.

Opioids are frequently administered epidurally because they augment analgesia with little respiratory compromise. 26Relatively little opioid is required in the epidural space compared with intravenous administration, and the dose we used was typical. 26,27Plasma fentanyl concentrations were six- to sevenfold lower after epidural than intravenous administration of the drug, which is consistent with the relatively small epidural dose. 28Furthermore, pupil size and reflex amplitude (which are excellent measures of opioid effect) were essentially unchanged by epidural fentanyl. As may be expected from these data, epidural fentanyl had no detectable influence on the febrile response.

We have previously demonstrated that opioid-induced inhibition of fever is mediated centrally, rather than by a reduction in peripheral concentrations of pyrogens, when opioids are given after IL-2. 9Our current results confirm this observation and extend it to the case in which opioids are given before induction of fever. Epidural analgesia did not inhibit fever and, not surprisingly, also failed to reduce circulating pyrogen concentrations.

We evaluated volunteers rather than patients. Responses under clinical circumstances are likely to differ somewhat. Nonetheless, it remains likely that the cause we propose for hyperthermia associated with epidural analgesia applies widely. Our theory does not exclude other causes of hyperthermia during epidural analgesia; it therefore remains likely that other yet-to-be-identified mechanisms also contribute.

In summary, the conventional assumption is that hyperthermia during epidural analgesia is caused by the technique. We tested an alternative theory that fever in humans is manifested normally during epidural analgesia, but suppressed by low-dose opioid. Intravenous fentanyl halved the febrile response to pyrogen, whereas epidural ropivacaine and epidural ropivacaine–fentanyl did not inhibit fever. These data support our proposed mechanism for hyperthermia during epidural analgesia and suggest that hyperthermia during epidural analgesia should thus be taken seriously and not considered a benign complication of the anesthetic technique per se .