Pharmacokinetics of tramadol after subcutaneous administration in a critically ill population and in a healthy cohort

Healthy male subjects aged 18 to 65 years (inclusive) were recruited by advertisement in the university community. Body weight was greater than 50 kg and body mass index (BMI) was between 19 and 30 kg/m². There was a requirement for adequate venous access in the left or right arm to allow collection of a number of blood samples. Health status was ascertained a week prior to the study by medical interview, physical examination and laboratory investigations including haematology, coagulation, biochemistry, liver function and urine drug screen (DipScan, Point of Care Diagnostics Australia Pty Ltd, Artarmon, NSW). All subjects were followed up one week after the study, when they repeated the medical interview and examination and reported any adverse events following the study.

Patients from the high dependency unit in a tertiary referral university hospital were considered for inclusion if they had been admitted with moderate to severe pain following surgery or trauma, were aged between 18 and 75 years of age, and were receiving fentanyl by patient-controlled analgesia (PCA) for management of their acute pain. All patients underwent detailed workup including history, physical examination and laboratory investigations including haematology, coagulation, biochemistry, liver function and urine drug screen (DipScan, Point of Care Diagnostics Australia Pty Ltd, Artarmon, NSW). Reliable venous access in the left or right arm was required to allow collection of a number of blood samples.

Patients were given fentanyl by PCA because it is metabolised by the cytochrome P450 enzymes CYP3A4/5 and not CYP2D6, the latter being the main enzyme responsible for the metabolism of tramadol and production of O-desmethyltramadol (M1), the active metabolite of tramadol responsible for most of the drug’s opioid agonist effect. Informed consent was obtained from the patients the day before the study commenced.

Exclusion criteria for both the healthy subjects and patients included an allergy or hypersensitivity to opioids, current use of opioid or psychoactive medications (including sedatives, hypnotics or tranquillisers), current or previous alcohol or drug abuse, current or recent use of medications or herbal preparations known to induce or inhibit CYP2D6/CYP3A4 enzymes, renal impairment (defined as a calculated creatinine clearance of ≤ 75 ml/min), donation of blood within three months prior to the study, or any condition that would interfere with blood sampling or drug disposition [18].

Exclusion criteria also included abnormal liver function (in healthy subjects defined as outside the normal laboratory range; in the patient group defined as liver function tests > 3 times the upper limit of normal range). Healthy subjects were also excluded if the urine drug screen was positive for any opioids, benzodiazepines, amphetamines, cocaine or cannabis.

The same protocol for tramadol administration and blood sampling was followed as for the healthy subjects. Acute Physiological and Chronic Health Evaluation (APACHE II) scores were calculated on the day of the study. The APACHE II score is a classification system designed to assess the severity of disease for adult patients admitted to an intensive care unit [19]. Urine was also collected cumulatively for the first 10 hours of the study for measurement of tramadol, M1 and M2 concentrations.

During the study period, all patients and healthy subjects had respiratory and heart rates, sedation scores, temperature and the severity of nausea/vomiting recorded every 30 minutes for the first 3 hours, and then hourly for the duration of the study. Level of sedation was scored on a 4-point scale from 0 to 3 (0 = wide awake, 1 = rouses easily and can stay awake; 2 = rouses easily but has difficulty staying awake, and 3 = somnolent, difficult to rouse) [5]. The degree of nausea and vomiting/retching was scored on a scale from 0 to 3 (0 = none, 1 = mild and does not need treatment, 2 = treatment was effective, 3 = persists despite treatment). Peripheral oxygen saturation (SpO₂) levels and respiratory rates were also monitored and the lowest values recorded. Any other adverse effects reported by the patients and healthy subjects were also documented.

The blood samples were stored in 10 ml heparinised plastic tubes and centrifuged within 2 hours of collection. Following centrifugation, the plasma was transferred into identically labeled polypropylene tubes. Samples were then frozen at -20°C and transferred to a -70°C freezer within 24 hours of processing. The samples were later transferred to the analytical laboratory and stored at -20°C for tramadol determination.

Plasma and urine tramadol, M1 and M2 concentrations were determined by high-performance liquid chromatography (HPLC) with fluorescence detection and using an external standardisation procedure as no internal standard was readily available at that time. This then required exact volumes of all liquids as outlined below. Plasma sample preparation was as follows: plasma (500 μl) was alkalinised with 1 M (100 μl) sodium hydroxide prior to extraction in 3 ml hexane: ethyl acetate (80:20). The organic layer (2.7 ml) was then back extracted into 0.05 M (150 μl) hydrochloric acid, the organic phase was aspirated and 100 μl of the remaining acidic phase was injected on to the HPLC system. Urine sample preparation was as follows: samples were centrifuged (6 minutes, 14000 r.p.m.), the supernatant was diluted to 1 in 100 in mobile phase (details below) and injected (100 μl) on to the HPLC system. The retention time of tramadol was 5.4 minutes (plasma) and in urine was 8.0 minutes and M1 and M2 metabolites were 3.7 and 8.7 minutes, respectively following mobile phase modification. Peak areas were used as output responses.

The HPLC system was a LC Workstation Class LC10 (Shimadzu, Kyoto, Japan) consisting of a SIL-10ADvp auto injector and LC-10ADvp liquid chromatograph (pump), with fluorescence detection (excitation and emission wavelengths of 210 and 305 nm, respectively; LC-240 Perkin Elmer, Buckinghamshire, UK) and a C-R6A Chromatopac integrator (Shimadzu, Kyoto, Japan). Tramadol, M1 and M2 were separated on a C18 LUNA analytical column (150 × 4.6 mm, Phenomenex, Lane Cove, Australia). The mobile phase for urine samples consisted of acetonitrile: 25 mM di-potassium hydrogen phosphate (39:61, v/v) adjusted to pH 8.9 with 85% orthophosphoric acid at a flow rate of 1.0 ml/min, and the mobile phase for plasma samples consisted of acetonitrile: 25 mM di-potassium hydrogen phosphate (15:75, v/v) adjusted to pH 3.0 with 85% orthophosphoric acid at a flow rate of 1.5 ml/min.

Calibration curves for tramadol, M1 and M2 quantification from urine samples were constructed with 6 final concentrations (ng/ml) ranging from 500 to 5000, 10 to 475 and 75 to 475, respectively. Calibration samples were prepared identically in blank human urine. Low, medium and high QC samples of tramadol, M1 and M2 were also prepared with final concentrations (ng/ml) of 750, 75 and 75, respectively; 2500, 250 and 250, respectively; and 4750, 475 and 475, respectively. The inter- and intra-assay imprecision and inaccuracy of the assays (n = 6) based on the QC samples were all < 10%. At the LLOQ of 25 ng/ml, intraday imprecision was 8.7% and inaccuracy was 4.2%; at the low (75 ng/ml), medium (200 ng/ml) and high (350 ng/ml) QCs, the values were 6.5% and 1.6%, 9.0% and 2.7% and 6.7% and 4.3%, respectively (all n = 6). For interday imprecision and inaccuracy, the values were 17.4% and 5.8%, 6.9% and 2.9%, 9.2% and 6.4% and 2.6% and 2.8% (n = 6). There was no extraction required for the urine samples and the recovery was >95%.

Calibration curves for tramadol quantification from plasma samples were constructed with 6 final concentrations ranging from 25 to 1000 ng/ml. Calibration samples were prepared identically in blank human plasma. Low, medium and high quality control (QC) samples of tramadol were also prepared with final concentrations (ng/ml) of 75, 200 and 350, respectively. Extraction recovery of tramadol was > 54.3%. The inter- and intra-assay (n = 6) imprecision and inaccuracy of the assays based on the QC analysis were all < 17.5%. The lower limit of quantification was 25 ng/ml. The protocols in both groups were exactly the same and blood samples were collected up to 24 hours after tramadol dosing. However, the curves were truncated because at certain time points, the concentrations fell below the limit of quantification (<LOQ) and were not easily detectable using the assay described (Figure 1).

The assay in both plasma and urine from patient samples revealed no additional peaks or shoulders on the chromatogram and chromatograms from other patient samples not administered tramadol contained no peaks at the retention times. Plasma and urine samples stored at -20°C for 3.6 and 9 months revealed no time-dependent changes in concentration.

The data were analysed using a non-compartmental approach (NCA) based on analysis of calculated and extrapolated area under the curve (AUC_0-∞) [20]. The AUC from zero time to the last quantifiable concentration was calculated by trapezoidal integration (linear up, log down method) and extrapolated to time infinity by assuming an exponential decline (best fit of 3 to 6 data points) to estimate the terminal exponential rate constant and therefore terminal half-life (t½). The analysis used previously validated scripts written in the R language [21]. Other pharmacokinetic parameters including time taken to reach maximum plasma concentrations (t_max), maximum plasma concentration (C_max), t½ and mean residence time (MRT) were also calculated from the observed data. Urine concentrations of tramadol, M1 and M2 metabolites and total urine volume were used to calculate the total amount of each excreted in 10 hours.

Demographic data are expressed as median (range) and were analysed using a two-tailed unpaired t-test. Pharmacokinetic data analysis used the R data analysis and statistical language (Version 2.10.1) using a single factor ANOVA [21]. Values were expressed as mean (±SD), or as mean (range) depending on their distribution.

The tramadol blood concentration-time curves for the individuals in each group are summarised in Figure 1 and the mean plasma concentration times are shown in Figure 2. There were no significant differences in AUC_0-∞ between the healthy subjects and the patients. The non-compartmental pharmacokinetic parameters are summarised in Table 2. In both cohorts, absorption of tramadol after SC administration was rapid. There were no significant differences in any of the pharmacokinetic variables (AUC_0-∞, p = 0.96; t_max, p = 0.73; C_max, p = 0.67; t½, p = 0.96; MRT, p = 0.97). The range of t_max values was 5–60 minutes in both the healthy subjects and the patients.

Urinary excretion of tramadol, M1 and M2 metabolites in both cohorts in our study were measured. They showed respective mean values of 28.2% tramadol, 4.01% M1 and 1.01% M2 in the urine collected over 10 hours in the healthy subjects. In the patient cohort, values were 24.3% tramadol, 3.43% M1 and 2.02% M2. The mean M1/tramadol excreted dose ratios were 0.14 ± 0.12 (range 0.01 to 0.34) for healthy subjects and 0.17 ± 0.14 (range 0.02 to 0.51) for the patient cohort. The mean M2/tramadol excreted dose ratios were 0.04 ± 0.03 (range 0.01 to 0.07) for healthy subjects and 0.08 ± 0.09 (range 0.06 to 0.33) for the patient cohort.

While subcutaneous tramadol would not be of interest in mechanically ventilated critically ill patients where intravenous sedation and analgesia is a requirement, subcutaneous pain relief can be useful in spontaneously ventilating patients in a high dependency setting [3, 4]. It may avoid the need for IV access and reduce the incidence of respiratory and gastrointestinal side effects when compared to IV administration [2]. However, the pharmacokinetics of drugs cannot be presumed to be the same in severely ill patients as in healthy subjects. Trauma, the perioperative state and critical illness may affect drug absorption and distribution due to differences in skin perfusion and cardiac output. In a similar study investigating the pharmacokinetics of oxycodone after a single SC dose, marked differences were revealed between healthy subjects and a similar cohort of patients [22]. The latter showed that AUC_0-∞, C_max and t½ were all significantly lower in the patient cohort despite no differences in t_max or MRT. In contrast, in this study of SC tramadol, no significant differences were found in AUC_0-∞ t½, t_max, C_max or MRT between patients and healthy subjects.

The SC pharmacokinetics of absorption of the following opioids has previously been reported: morphine in healthy subjects and older patients; fentanyl in healthy subjects; and oxycodone in both healthy subjects and critically ill patients [22, 31‐33]. The mean t_max values for tramadol in this study of 20.6 and 23.3 minutes in healthy subjects and patients, respectively, were comparable to those reported for oxycodone in both healthy subjects and critically ill patients (22.1 and 20.5 minutes respectively), but longer than that reported for fentanyl and morphine in subjects (both 15 minutes) or morphine in elderly patients (15.9 minutes).

In comparing the pharmacokinetic parameters of tramadol after oral, SC, IV and IM routes of administration in healthy subjects, our study using the SC route of 50 mg tramadol showed that the t_max was 20.6 minutes, t½ was 5.2 hours, C_max was 0.47 mcg/ml and AUC was 175 mcg/ml*min. Previous reports have shown that a 100 mg oral dose of tramadol in healthy subjects gave a t_max value of 66 minutes, a t½ value of 5.6 hours, a C_max value of 0.31 mcg/ml and an AUC value of 160 mcg/ml*min, whilst a single IV dose of 50 mg tramadol gave a t½ value of 5.5 hours, a C_max value of 0.35 mcg/ml and an AUC of 93 mcg/ml*min. With a single IM dose of 50 mg tramadol, the t_max was 45 minutes, t½ was 5.5 hours, C_max was 0.19 mcg/ml and the AUC was 95 mcg/ml*min [9].

There are limited data on the urinary excretion of tramadol and its metabolites. In healthy subjects given a 50 mg oral dose of tramadol, the mean values for urinary excretion over 24 hours (expressed as a percentage of the dose administered) were tramadol 12%, M1 15%, and M2 4% [34]. Following a 100 mg dose of oral tramadol, also given orally to healthy subjects, with urine collected over 30 hours, the mean values were 16.2% unchanged tramadol, 11.2% M1, and 1.1% M2 [35]. Similar results were reported by Rudaz et al. also following administration of 100 mg oral tramadol to healthy subjects, again with urine collected over 30 hours: 16% unchanged tramadol, 16% M1 and 2.2% M2 [36]. In our study, respective mean values were 28.2% tramadol, 4.01% M1 and 1.01% M2 in the urine collected over 10 hours in the healthy group. In the patient cohort, values were 24.3% tramadol, 3.43% M1 and 2.02% M2. Since the clearance via the M1 and M2 pathways does not appear to be different between the two groups (as assessed by the urinary metabolite to tramadol ratio), this suggests that the CYP2D6 (M1) and CYP3A4 (M2) pathways were not altered in severely ill patients.

The study was designed to avoid pharmacokinetic interactions with other drugs. In the patients fentanyl was chosen as the “rescue” opioid, delivered by PCA, for intercurrent pain relief because it is metabolised by the cytochrome P450 enzyme CYP3A4. CYP3A4 is also responsible for metabolism of tramadol to M2. However, fentanyl is only a substrate and not an inducer/inhibitor of CYP3A4 at therapeutic concentrations. If fentanyl was competing with tramadol as a substrate for this enzyme, the AUC_0-∞ results for tramadol would have been higher in the patients who were receiving fentanyl. Two of the patients received oral oxycodone during the study period. Oxycodone is metabolised by CYP3A4 and CYP2D6. Similarly, oxycodone is only a substrate and not known to be an inhibitor of CYP2D6, and therefore the results from these patients were not excluded.

A possible limitation of this study is the mismatching between the group in terms of age and sex. However given that no difference was noted in the pharmacokinetics between the groups it is unlikely that the results of the study would have been different if the groups had been better matched. Based on the paucity of pharmacokinetic data after SC administration of this drug, no a priori power calculation was performed, which is therefore one of the limitations of this study. Hence, between-group differences in some pharmacokinetic parameters, found to be similar in this study, cannot be confidently excluded as the study may not have been adequately powered to demonstrate a difference. Nevertheless, there is no trend for a difference in any parameter so any possible difference is likely to be small and may not be clinically relevant. The estimated effect size and its standard error are given in Table 2 to indicate the study power. Moreover, there may be some patient groups where the time-course of drug concentrations and effect are clinically different from the patients in the current study. We chose not to phenotype our subjects for CYP2D6, because total subject numbers would have been too small to detect significant differences.