A review of factors explaining variability in fentanyl pharmacokinetics; focus on implications for cancer patients

Abstract

Fentanyl is a strong opioid that is available for various administration routes, and which is widely used to treat cancer‐related pain. Many factors influence the fentanyl pharmacokinetics leading to a wide inter‐ and intrapatient variability. This systematic review summarizes multiple studied factors that potentially influence fentanyl pharmacokinetics with a focus on implications for cancer patients. The use of CYP3A4 inhibitors and inducers, impaired liver function, and heating of the patch potentially influence fentanyl pharmacokinetics in a clinically relevant way. In elderly patients, current data suggest that we should carefully dose fentanyl due to alterations in absorption and metabolism. The influence of BMI and gender on fentanyl pharmacokinetics is questionable, most probably due to a large heterogeneity in the published studies. Pharmacogenetics, e.g. the CYP3A5*3 gene polymorphism, may influence fentanyl pharmacokinetics as well, although further study is warranted. Several other factors have been studied but did not show significant and clinically relevant effects on fentanyl pharmacokinetics. Unfortunately, most of the published papers that studied factors influencing fentanyl pharmacokinetics describe healthy volunteers instead of cancer patients. Results from the studies in volunteers may not be simply extrapolated to cancer patients because of multiple confounding factors. To handle fentanyl treatment in a population of cancer patients, it is essential that physicians recognize factors that influence fentanyl pharmacokinetics, thereby preventing potential side‐effects and increasing its efficacy.

Tables of Links

TARGETS
Enzymes 2	Transporters 3
CYP3A4	ABCB1
CYP3A5	SLCO1B1

LIGANDS
Carbamazepine	Fentanyl
Haloperidol	Ketoconazole
Morphine	Parecoxib
Phenobarbital	Rifampicine

These Tables list key protein targets and ligands in this article that are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY 1, and are permanently archived in the Concise Guide to PHARMACOLOGY 2015/16 2, 3.

Introduction

Pain is a common and relevant problem in cancer patients 4. Chronic pain occurs in about 30–50% of cancer patients undergoing curatively aimed treatment and in 70–90% of patients with advanced disease 5. Breakthrough pain is found in 64–90% of cancer patients with chronic pain 6. It is defined as a ‘transitory exacerbation of pain experienced by the patient who has relatively stable and adequately controlled baseline pain’ 7, 8. Moderate to severe cancer pain is commonly treated with opioids, of which fentanyl is one of the most widely used, especially because it is available for various easy‐to‐use administration routes. Since the 1990s the fentanyl transdermal patch has been used to treat chronic pain 9. Nowadays, there are also several rapid onset products available for the treatment of breakthrough pain: products for transmucosal, buccal, sublingual, and even intranasal administration 10.

Fentanyl is a strong opioid (approximately 75–100 times more potent than morphine), highly lipophilic and binds strongly to plasma proteins 11, 12. Its volume of distribution is large (3.5–8 l kg⁻¹) and its clearance relatively high (30–72 l h⁻¹) 12. Fentanyl is thought to be predominantly metabolized in the liver by cytochrome P450 iso‐enzyme 3A4 (CYP3A4)‐mediated N‐dealkylation, resulting in the inactive metabolite norfentanyl. Less than 1% is metabolized by alkyl hydroxylation, N‐dealkylation or amide hydrolysis to the inactive compounds hydroxyfentanyl, hydroxynorfentanyl and despropionylfentanyl. The inactive metabolites, and approximately 10% of the intact molecule, are mainly excreted by the kidneys 13, 14, 15 (Figure 1). Although this is the most accepted explanation of fentanyl metabolism, a recent study showed that the CYP3A4‐mediated N‐dealkylation step may not be as important as previously thought. Unknown metabolic routes may be responsible for a significant part of fentanyl metabolism 16.

Figure 1.

Fentanyl metabolism and elimination 73

When fentanyl is given as an intravenous bolus, fentanyl is rapidly distributed from plasma into highly vascularized compartments. After uptake in the systemic circulation, redistribution to muscle and fat tissue occurs. Elimination half time is highly variable in various studies (219–853 min), particularly due to this redistribution 17, 18, 19, 20. Fentanyl used in a transdermal patch is absorbed first by the skin, then taken up into the cutaneous microcirculation, before entering the systemic circulation. The rapid‐onset fentanyl products are absorbed by the highly vascularized oromucosal and nasal membranes before entering the systemic circulation. Additionally, there is also some gastrointestinal uptake, especially for the oromucosal products 10.

There exists a wide inter‐individual and inter‐occasion (= intra‐individual) variability (IIV and IOV) in the pharmacokinetics of fentanyl 14, 21, 22, 23, 24, 25. Several factors may cause variability by influencing absorption, distribution, metabolism and/or excretion (ADME) 26, 27, 28. The intravenous route of administration bypasses the absorption step and is consequently not influenced by factors that affect the absorption rate. This explains the relatively high IIV and IOV of fentanyl given by the transdermal, oromucosal and nasal routes compared to intravenously administered fentanyl 21, 29. Although fentanyl is mostly dosed by titration, nevertheless in specific situations, such as a rotation to fentanyl from another opioid or a change in co‐medication, under‐ or overdosing may occur. Therefore knowledge of factors that influence the variability of fentanyl pharmacokinetics is important. This review aims to study the factors that cause variation in fentanyl pharmacokinetics to increase the understanding of fentanyl pharmacokinetics and its safe use in cancer‐related pain. To better understand the mechanism by which the various factors influence pharmacokinetics, we separately describe studies using intravenous fentanyl and studies using other administration routes.

Methods

We performed a review on factors related to pharmacokinetic aspects of immediate release and slow release fentanyl preparations. We searched in PubMed, Cochrane and Embase (supplement A). The main (Mesh) terms we used were: (fentanyl), (intravenous), (cutaneous), (transdermal), (sublingual), (buccal), (nasal), (transmucosal), (oral), (pharmacokinetics), (biotransformation), (tissue distribution) and (elimination). The search was limited to English or Dutch articles published until July 2014, followed by an update until January 2016.

Additional papers were found by searching the references in selected articles for cross‐references. The articles were independently reviewed for eligibility by two authors (EJMK and MLZ). Studies were included in the analysis if they contained pharmacokinetic parameters (e.g. area under the curve (AUC), Clearance (CL), Time to maximum concentration (T _max), plasma concentrations), described regularly available fentanyl products and were published in English or Dutch. When the same cohort of patients was described in more than one paper, the paper fitting the inclusion criteria best was chosen. Exclusion criteria were: full text not available, no original research reported, studies in populations other than adults, fentanyl not being the main subject studied, no factors studied in relation to pharmacokinetic parameters, no standard administration route of fentanyl and no pharmacokinetics of fentanyl described. For every publication, we reviewed the type of study, the number of evaluable patients, the administration route, the studied population (e.g. healthy volunteers, cancer patients or peri‐operative patients), the number of pharmacokinetic (PK) samples taken per individual (Table 1), blood sample analysis or patch residue analysis, the pharmacokinetic parameters calculated, studied covariate(s) (e.g. age, gender), and the size of the measured effect(s). Drug/molecular target nomenclature is used according to the Concise Guide to Pharmacology 30.

Table 1.

Overview of the characteristics of included studies

Author	Reference	Year	Country	Study type	N	Administration route	Patients	Number of PK samples per patient
Ariano	53	2001	Canada	Non‐randomized clinical trial	18	iv	Healthy volunteers	14 samples
Ashburn	48	2003	United States	Open randomized cross over study	5, 7 and 9	Transdermal	Healthy volunteers	>36 samples
Barratt	21	2014	Europe (11 countries) cohort	Cross‐sectional study	620	Trandermal	Cancer patients	1 sample
Bentley	29	1982	United States	Non‐randomized clinical trial	9	iv	Post operative patients	19 samples
Darwish	63	2007	United States	Non‐randomized clinical trial	16	Buccal	Cancer patients	13 samples
Finn	64	2011	United States	Non‐randomized clinical trial	14	Buccal	Cancer patients	8 samples
Gupta	51	1995	United States	Non‐randomized clinical trial	6, 8 and 11	iv	Healthy volunteers	12,27,51 samples
Han	67	2007	Korea	Non‐randomized clinical trial	20	iv	Patients with burns	20 samples
Heiskanen	62	2009	Finland	Non‐randomized clinical trial	20	Transdermal	Cancer patients	5 samples
Holdsworth	54	1994	United States	Non‐randomized clinical trial	16	Transdermal	Healthy volunteers	47 samples
Ibrahim	33	2003	United States	Randomized crossover study	12	iv	Healthy volunteers	20 samples per test
Kharasch	55	2004	United States	Non‐randomized clinical trial	24	Transmucosal	Healthy volunteers	17 samples
Kharasch	15	2004	United States	Randomized crossover study	12	Transmucosal	Healthy volunteers	17 samples
Koehntop	36	1997	United States	Non‐randomized clinical trial	8	iv	Patients undergoing renal transplantation	14 samples
Kokubun	37	2012	Japan	Cross‐sectional study	51	Transdermal	Cancer patients	3 samples
Moore	49	2012	United States	Randomized cross over study	20	Transdermal	Healthy volunteers	18 samples
Nomura	61	2013	Japan	Non‐randomized clinical trial	18	Transdermal	Cancer patients	8 samples
Olkkola	26	1999	Finland	Randomized double blind placebo controlled cross over study	11	iv	Healthy volunteers	15 samples
Palkama	34	1998	Finland	Randomized double blind cross over study	10	iv	Healthy volunteers	14 samples
Parikh	50	2013	United States	Randomized crossover study	29	Sublingual	Healthy volunteers	18 samples
Perelman	65	2013	Canada	Randomized crossover study	31	Nasal	Rhinitis	16 samples
Saari	27	2008	Finland	Randomized crossover study	12	iv	Healthy volunteers	14 samples
Shomaker	28	2000	United States	Non‐randomized cross over study	6	Transdermal	Healthy volunteers	25 samples
Singleton	56	1988	United States	Non‐randomized clinical trial	14	iv	Post operative patients	19 samples
Solassol	44	2005	France	Observational study	108 (507 patches)	Transdermal	Cancer patients	N.A.
Solassol	57	2005	France	Non‐randomized clinical trial	29	Transdermal	Cancer patients	2–4 samples
Takashina	68	2012	Japan	Non‐randomized clinical trial	60	Transdermal	Cancer patients	1 sample
Tanaka	69	2014	Japan	Non‐randomized clinical trial	52	iv	After surgery	1 sample
Thompson	58	1998	United Kingdom	Non‐randomized clinical trial	18	Transdermal	Post operative patients	18 samples
Van Nimmen	39	2010	Belgium	Observational study	68 (498 patches)	Transdermal	Cancer patients	N.A.
Ziesenitz	38	2013	Germany	Prospective randomized crossover study	16	iv	Healthy volunteers	20 samples per test
Ziesenitz	16	2015	Germany	Randomized crossover study	16	iv	Healthy volunteers	20 samples

Results

Results of the search

The original search of Pubmed, Embase and Cochrane resulted in 1543 citations; one extra article was found by checking cross references. Of these 1544 papers, 31 papers met all the inclusion criteria. The recent update led to one additional paper 16 meeting the strict inclusion criteria, so in total 32 papers were taken into account (see Figure 2 and Table 1). The majority of the studies were performed in healthy volunteers (n = 14), about a third in cancer patients (n = 11) and the remaining seven studies in selected other patient populations (patients studied during and after elective surgery, during renal transplantation, patients with rhinitis, and patients with burns). In 13 studies transdermal or intravenous fentanyl was used and in six studies oromucosal fentanyl was used.

Figure 2.

Flow chart of study sample inclusion/exclusion

In total, 36 related factors were investigated in these studies. We divided the related factors into four groups: pharmacokinetic drug–drug interactions (Table 2), environmental factors (Table 2), patient‐related factors (Table 3) and pharmacogenetics (Table 4).

Table 2.

Pharmacokinetic drug interactions & environmental factors

Factor	Administration route	N	Study group	Blood‐samples (BS) or residue in patch analysis (PA)	AUC	C max	T max	CL	Cld	t 1/2	Vd	Absorption from patch	Other	Significant effect	Reference
CYP3A4 inhibitors
Troleandomycin	OTFC	12	Healthy volunteers	BS	↑	X	X						Kel, V/F, CL/F	AUC0–∞ (10.4 ± 8.9 h*ng ml−1) with troleandomycin AUC0–∞ (5.87 ± 3.74 h*ng ml−1) without troleandomycin	Kharasch 15
	IV	12	Healthy volunteers	BS	↑	X	X	↓		X	X		Kel ↓	AUC0–∞ (9.94 ± 3.77 h*ng ml−1) with troleandomycin AUC0–∞ (6.04 ± 2.19 h*ng ml−1) without troleandomycin	Ibrahim 33
Ritonavir	IV	11	Healthy volunteers	BS	↑			↓		↑	X			AUC0–∞ (18.1 ± 6.5 h*ng ml−1) with ritonavir AUC0–∞ (6.6 ± 3.4 h*ng ml−1) without ritonavir CL 5.2 ± 2.0 ml min−1 kg−1 with ritonavir CL 15.6 ± 8.2 ml min−1 kg−1 without ritonavir t 1/2: 20.1 ± 8.4 h with ritonavir t 1/2: 9.4 ± 4.6 h without ritonavir (P < 0.01)	Olkkola 26
Voriconazole	IV	12	Healthy volunteers	BS	↑			↓		X	X			AUC0–∞ (8.5 ± 2.9 h*ng ml−1) with voriconazole AUC0–∞ (6.1 ± 1.1 h*ng ml−1) without voriconazole CL 10.7 ± 3.0 ml min−1 kg−1 with voriconazole CL 14.0 ± 2.5 ml min−1 kg−1 without voriconazole	Saari 27
Ketoconazole	IV	16	Healthy volunteers	BS	↑	X		↓		X	X			AUC0–∞ (6.8 ± 3.4 h*ng ml−1) with ketaconazole AUC0–∞ (5.1 ± 2.5 h*ng ml−1) without ketaconazole CL 14.7 ± 5.6 ml min−1 kg−1 with ketaconazole CL 19.0 ± 6.8 ml min−1 kg−1 without ketaconazole	Ziesenitz 16
Itraconazole	IV	10	Healthy volunteers	BS				X		X	X			No	Palkama 34
Fluconazole	IV	12	Healthy volunteers	BS	X			↓		X	X			CL 11.6 ± 3.0 ml min−1 kg−1 with fluconazole CL 14.0 ± 2.5 ml min−1 kg−1 without fluconazole	Saari 27
Grapefruit juice	OTFC	12	Healthy volunteers	BS	X	X	X						Kel, V/F, CL/F	No	Kharasch 15
CYP3A4 inducers
Rifampicin	OTFC	12	Healthy volunteers	BS	↓	X	X						Kel ↑, V/F, CL/F ↑	AUC0–∞ (2.20 ± 0.84) with rifampicin AUC0–∞ (5.87 ± 3.74) without rifampicin	Kharasch 15
Carbamazepine or phenobarbital	Patch	51	Cancer patients	BS	X			↑			X		Ka, tL	Significant influence on CL ‐> NONMEM analysis CLfenta (L/h) = 3.53 × (15‐CPS) × (1 + 1.38 × *) * CYP3A4 inducers = 1, no CYP3A4 inducers = 0	Kokubun 37
Other medication
Parecoxib	IV	12	Healthy volunteers	BS	X	X	X	X		X	X		Kel	No	Ibrahim 33
Haloperidol	Patch	68 (498 patches)	Cancer patients	PA								X	Urinary elimination	No	V Nimmen 39
Morphine	Patch	68 (498 patches)	Cancer patients	PA								X	Urinary elimination	No	V Nimmen 39
Localization of the patch	Patch	108 (507 patches)	Cancer patients	PA								X		No	Solassol 44
	Patch	68 (498 patches)	Cancer patients	PA								↑	Urinary elimination	Overall significant influence of site of application (arm, torso, leg) of the patch on transdermal fentanyl delivery (P = 0.0011); 7.5% higher delivery efficiency at the arm compared to the leg	V Nimmen 39
Local heat on patch	Patch	5,7,9	Healthy volunteers	BS	↑	↑	X							AUC0–4 h 1.22 ± 0.37 ng*h ml−1 with heat AUC0–4 h 0.42 ± 0.35 ng*h ml−1 without heat C max 0.63 ± 0.15 ng ml−1 with heat (after 4 h) C max 0.24 ± 0.20 ng ml−1 without heat (after 4 h)	Ashburn 48
	Patch	6	Healthy volunteers	BS	↑	↑	X							AUC0–4 h 39.1 (9.6–76.8) ng*h ml−1 with heat AUC0–4 h 11.3 (0.1–18.0) ng*h ml−1 without heat C max 0.397 (0.14–0.69) ng/ml−1 with heat (after 4 h) C max 0.126 (0.00–0.18) ng ml−1 without heat	Shomaker 28
	Patch	20	Healthy volunteers	BS	↑	↑								AUC0–10 h 1.7 ± 0.9 ng*h ml−1 with heat AUC0–10 h 0.7 ± 0.5 ng*h ml−1 without heat C max 0.5 ± 0.3 ng ml−1 with heat (after 10 h) C max 0.3 ± 0.2 ng ml−1 without heat (after 10 h)	Moore 49
Hot/cold beverages	SL	29	Healthy volunteers	BS	X	X	X			X			λz	No	Parikh 50
Other factors
Low/high pH beverages	SL	29	Healthy volunteers	BS	X	X	X			X			λz	No	Parikh 50
Alcohol consumption	Patch (chronic use)	108 (507 patches)	Cancer patients	PA								X		No	Solassol 44
Smoking	Patch (chronic use)	108 (507 patches)	Cancer patients	PA								X		No	Solassol 44
Diurnal variation	IV	6, 8, 11	Healthy volunteers	BS	X									No	Gupta 51

AUC, area under the curve; CL, clearance; CLd, distributional clearance; CL/F, apparent oral clearance; C _max, maximum concentration; ka, absorption rate constant; IV, intravenous; Kel, terminal elimination rate constant; NONMEM, nonlinear mixed‐effect model; OTFC, oral transmucosal fentanyl citrate; SL, sublingual; t _L, lag time; T _max, time to reach maximum concentration; t _1/2, half‐life; Vd, volume of distribution; V/F, apparent volume of distribution; X, factor is studied but result not significant; λ_z, terminal disposition rate constant

Table 3.

Patient‐related factors

Factor	Administration route	N	Study group	Blood samples (BS) or residue in patch analysis (PA)	AUC	C max	T max	CL	Cld	t 1/2	Vd	Absorp‐tion from patch	Other	Significant effect	Reference
Age	IV	18	Healthy volunteers	BS				X	↑		Xa			Higher CLd in elderly (mean 14.59 l kg−1 h−1 vs. 3.18 l kg−1 h−1)	Ariano 53
	IV	9	Perioperative patiens	BS				↓		↑	X			Longer t 1/2 in pt >60 yr vs. pt <50 yr (945 vs. 265 min). Lower CL in pt >60 yr vs. pt <50 yr (265 ml min−1 vs. 991 ml min−1).	Bentley 29
	IV	14	Perioperative patients	BS				X		X	↓		Plasma fentanyl concentration ↑	Higher plasma concentration 2 min (mean 7.73 ng ml−1 vs. 4.54 ng ml−1) and 4 min (mean 3.26 ng ml−1 vs. 1.78 ng ml−1) after infusion in elderly (71–82 yr) vs. younger(18–41 yr). Lower VDss in elderly (mean 1.36 l kg−1 in vs. 2.27 l kg−1)	Singleton 56
	OTFC	12	Healthy volunteers	BS	X	X	X						2nd peak T max, Kel (1/h), CL/F (l/h), V/F (l)	2nd peak T max 1.3 h ± 0.5 (elderly) vs. 1.9 ± 0.5 (young)	Kharasch 55
	Patch	16	Healthy volunteers	BS	↑	X								Mean AUC0–60 h/PD 2.05 ng ml−1 (67–87 yr) vs. 0.88 ng ml−1 (19–27 yr)	Holdsworth 54
	Patch (first)	18	Peri/post‐operative Patients	BS	X	X	X			X			Halftime ↑	Halftime; in pt 64–82 yr vs. pt 25–38 yr (11.1 h vs. 4.2 h)	Thompson 58
	Patch (chronic use)	108 (507 patches)	Cancer patients	PA								↓		Univariate analysis: Mean fentanyl absorption 55.7% (>75 yr), 63.7% (65–75 yr), 66 % (<65 yr)	Solassol 44
	Patch	51	Cancer patients	BS	X			X			X		Ka, NONMEM analysis	No	Kokubun 37
	Patch	68 (498 patches)	Cancer patients	PA								X	Urinary elimination	No	V Nimmen 39
	Patch	29	Cancer patients	BS/PA				Xb				X	Plasma concentration at steady state	No	Solassol 57
Gender	Patch	620	Cancer patients	BS									Serum fentanyl ↓and norfentanyl concen‐trations, MR	Serum fentanyl concentrations ↓ in men, however less than 1% of the variability	Barratt 21
	Patch	68 (498 patches)	Cancer patients	PA								X	Urinary elimina‐tion	Fentanyl excretion ↑ men / ↓ women (P = 0.04)	V Nimmen 39
	Patch	108 (507 patches)	Cancer patients	PA								X		No	Solassol 44
Albumin	Patch (first, after IV)	18	Cancer patients	BS									Dose‐adjusted serum fentanyl concentration ↓	Dose‐adjusted serum fentanyl concentration (mean) alb < 3.5 g dl−1–alb > 3.5 g dl−1 15 h 0.014–0.034 18 h 0.019–0.028 24 h 0.018–0.029	Nomura 61
	Patch	620	Cancer patients	BS									Serum fentanyl and norfentanyl concen‐trations, MR ↓	Inverse association between albumin and serum fentanyl concentrations, explained variability less than 1%. Negative association with MR, explained variability less than 1%.	Barratt 21
BMI	Patch	20	Cancer patients	BS									Dose adjusted plasma fentanyl concentration ↓	Cachexia (mean BMI 16 kg m–2) vs. normal weight (mean BMI 23 kg m–2): At 48 h dose‐adjusted plasma fentanyl concentration 0.014 μg l−1 vs. 0.023 μg l−1. At 72 h: dose‐adjusted plasma fentanyl concentration 0.012 μg l−1 vs. 0.024 μg l−1.	Heiskanen 62
	Patch	108 (507 patches)	Cancer patients	PA								X		No	Solassol 44
	Patch	68 (498 patches)	Cancer patients	PA								X	Urinary elimination	No	V Nimmen 39
	Patch (first, after IV)	18	Cancer patients	BS									Dose adjusted serum fentanyl concentrations	No	Nomura 61
	Patch	620	Cancer patients	BS									Serum fentanyl and norfen‐tanyl concen‐trations, MR ↓	Inverse association between BMI and serum fentanyl concentrations, explained variability less than 1%. Negative association with MR, explained variability less than 1%.	Barratt 21
	Patch	29	Cancer patients	BS/PA								X	Plasma concen tration at steady state	No	Solassol 57
Renal function
	IV	8	Patients undergoing renal transplantation	BS				↑			X		t 1/2 αβ, Vc, Ke, Kcp, Kpc	BUN > 60 mg dl−1 mean clearance 3.3 ml kg−1 min−1, BUN < 49 mg dl−1 mean clearance 11.7 ml kg−1 min−1	Koehntop 36
	Patch	620	Cancer patients	BS									Serum fentanyl and norfen tanyl concen‐trations, MR ↑	↑ MR, explained variability less than 2 %.	Barratt 21
	Patch	68 (498 patches)	Cancer patients	PA								X	urinary elimination	No	V Nimmen 39
Liver function
	Patch	51	Cancer patients	BS	↑			↓			X			CPS gr B ‐> AUC 1.36× larger than CPS gr A CPS gr C ‐> AUC 3.72× larger than CPS gr A CLfenta (l h−1) = 3.53 × (15 − CPS) × (1 + 1.38 × *) * CYP3A4 inducers = 1, no CYP3A4 inducers = 0	Kokubun 37
	Patch	68 (498 patches)	Cancer patients	PA								X	urinary elimination ↓	↓ urinary elimination in more severe liver function (P = 0.04)	V Nimmen 39
Mucositis	Buccal	16	Cancer patients	BS	X	X	X							No	Darwish 63
	Buccal	14	Cancer patients	BS	X	X	X							No	Finn 64
Rhinitis ± treated with oxymethalozine	Intranasal	31	Pts with rhinitis	BS	X	X	X			X				T max (min) median (range) Control 17 (10–120) Rhinitis 20 (5–180) Treated 53 (5–180) (oxymethalozine)	Perelman 65
Hypertrichosis	Patch (chronic use)	507 patches	Cancer patients	PA								X		No	Solassol 44
Hyperhidrosis	Patch (chronic use)	507 patches	Cancer patients	PA								X		No	Solassol 44
Burns	IV	20	Pts with burns	BS				↑		X			V	Burned: CL 29.4 ml min−1 kg−1; Unburned 21.0 ml min−1 kg−1	Han 67

AUC, area under the curve; AUC/PD, AUC/patch duration; BUN, blood urea nitrogen; CL, clearance; Cld, distributional clearance; CL/F, apparent oral clearance; C _max, maximum concentration; CPS, Child Pugh Score; Half‐time, time for plasma concentrations to double after patch application; IV, intravenous; ka, absorption rate constant; Kel, terminal elimination rate constant; Ke, elimination rate constant; Kcp, central to peripheral inter compartmental rate constant; Kpc, peripheral to central inter compartmental rate constant; MR, serum norfentanyl concentration/serum fentanyl concentration; NONMEM, nonlinear mixed‐effect model; t _L, lag time; T _max, time to reach maximum concentration; t _1/2, half‐life; Vc, volume of the central compartment; Vd, volume of distribution; V/F, apparent volume of distribution; X, factor is studied but result not significant.

a Vd is specified here as volume of distribution of the peripheral and central compartment

bCL = k0/Css where k0 is the perfusion rate computed from the true amount of fentanyl absorbed over 72 h and Css is the steady‐state fentanyl concentration

Table 4.

Pharmacogenetics

Factor	Administration route	N	Study group	Blood‐samples (BS) or residue in patch analysis (PA)	AUC	C max	T max	CL	Cld	t 1/2	Vd	Absorption from patch	Other	Significant effect	Reference
CYP3A5*3	Patch	60	Cancer patients	BS				↓					PCnMAR ↑, MAR, PC,	PCnMAR: *3/*3 group: 2.01 (1.21–2.44);*1/*1 group: 0.82 (0.77–1.25); *1/*3 group: 1.03 (0.80–1.74) ng ml−1 per μg h−1 kg−1 (P = 0.048 and P = 0.021). CL/F: CYP3A5*3/*3 group 28.0 (20.6–55.1); *1/*1 group 67.4 (47.2–78.2); *1/*3 group 58.4 (41.3–75.6) l h−1.	Takashina 68
CYP3A5*3 CYP3A5*1 carriers	IV	52	Postoperative patients	BS									nPC ↑, CLc, CLr, PC	nPC: *3/*3 group: 2.30 (1.09–2.98); *1 group: 1.21 (0.78–1.90) ng ml−1 per μg h−1 kg−1	Tanaka 69
ABCB1 C1236T	Patch	60	Cancer patients	BS									MAR, PC, PCnMAR	no	Takashina 68
SLCO1B1*a1 (genetic wildtype) or *15 (deficient haplotype with altered transport activitiy)	IV	16	Healthy volunteers	BS	X	X		X					AeF, CLren	no	Ziesenitz 38
CYP3A4*22 and CYP3A5*3	Patch	620	Cancer patients	BS									Serum fentanyl and norfenta‐nyl concen‐trations, MR	no	Barratt 21

AUC, area under the curve; AEf, amount of fentanyl excreted in urine within 24 h after administration; CL, clearance; Cld, distributional clearance; CL_ren, renal clearance; CL/F, total clearance of fentanyl; CL_c, creatine clearance (urine); CL_r,renal clearance (urine); MAR, measured absorption rate; MR, serum norfentanyl concentration/serum fentanyl concentration; nPC, plasma concentrations of fentanyl normalized for infusion rate; PC, plasma concentration of fentanyl; PCnMAR, plasma concentration of fentanyl normalized for the measured absorption rate; t _1/2, half‐life; X, factor is studied but result not significant.

Pharmacokinetic drug–drug interactions

Pharmacokinetic drug–drug interactions are highly relevant in daily clinical practice 31. Especially cancer patients frequently use numerous concomitant drugs during the different phases of their disease. In particular, CYP3A4 inhibitors or inducers may influence fentanyl pharmacokinetics because this iso‐enzyme is involved in the conversion of many drugs and also fentanyl is predominantly metabolized by CYP3A4. CYP3A4 inhibitors are divided into strong, moderate and weak inhibitors according to their pharmacokinetic effects. By definition, a strong CYP3A4 inhibitor results in a >5‐fold increase in the plasma AUC of a sensitive CYP3A4 substrate and a strong CYP3A4 inducer results in a >80% increase in clearance. Moderate inhibitors cause a >2‐fold increase in AUC or a decrease of clearance by 50–80%; and weak inhibitors still cause a clinically relevant >1.25‐fold but <2‐fold increase in AUC or a decrease of clearance by 20–50% 32. Meanwhile, strong CYP3A4 inducers may lead to a (large) reduction in plasma drug concentrations.

CYP3A4 inhibitors

The influence of several strong CYP3A4 inhibitors on fentanyl pharmacokinetics has only been studied in healthy volunteers so far. Most studies were performed with intravenous fentanyl and two studies with transmucosal fentanyl. The largest effects were found in volunteers receiving the antiretroviral drug ritonavir and the anti‐mycotic compound troleandomycine 15, 33. Concurrent use of ritonavir and (intravenous) fentanyl led to an 2.7‐fold increase in the AUC of fentanyl, concurrent use of troleandomycine and (intravenous and transmucosal) fentanyl led to an almost 2‐fold increase in the AUC of fentanyl, both drugs compared to fentanyl alone. This effect was similar for the intravenous and transmucosal route of fentanyl, although the standard deviation (SD) of the AUC in the transmucosal group doubled the SD in the group with intravenous fentanyl, reflecting the variable absorption rates for transmucosal fentanyl 15, 33. Use of voriconazole and ketoconazole in combination with fentanyl led to slightly smaller effects on the fentanyl AUC (mean increased AUC of 1.4 times and 1.3 times, respectively) 16, 27. Use of itraconazole showed a non‐significant 9% decrease in fentanyl clearance 34. As expected, with the use of moderate CYP3A4 inhibitors, effects were less relevant. Fluconazole in combination with intravenous fentanyl gave a slightly decreased plasma clearance of fentanyl but no significant differences in AUC 27. In another study, grapefruit juice, a strong CYP3A inhibitor, did not lead to changes in pharmacokinetic parameters when combined with transmucosal fentanyl 15.

Interestingly, in all studies, the measured effects on fentanyl pharmacokinetics were relatively small compared to other CYP3A substrates. A pharmacological explanation for some of the small drug–drug effects may be the high extraction ratio of fentanyl. The plasma clearance of fentanyl almost equals hepatic blood flow. For drugs with a high extraction ratio, like fentanyl, a variation of the intrinsic ability to eliminate a drug results in only marginal change in total clearance. In this case, clearance is affected mainly by liver blood flow 35, 36. However, Ziesenitz et al. hypothesized unknown metabolic pathways to explain relatively small effects of ketoconazole on the AUC of fentanyl. They showed a marked inhibition of the formation of norfentanyl with ketoconazole, while the clearance of fentanyl was only modestly decreased. Furthermore, measurements of fentanyl and known metabolites in urine could not retrieve the dose of fentanyl that was administered 16.

CYP3A4 inducers

Also the CYP3A4 inducers rifampicin, carbamazepine and phenobarbital were studied for their influence on fentanyl pharmacokinetics 15, 37, 38. None of these studies was performed with intravenously administered fentanyl. Administration of rifampicin to volunteers using transmucosal fentanyl led to a 2.6‐fold lower AUC compared to fentanyl alone 15. Carbamazepine and phenobarbital, the used CYP3A4 inducers in a study by Kokubun et al., led to a significantly higher clearance (>2× as high) of transdermal fentanyl compared to patients not using CYP3A4 37.

Other

Parecoxib and haloperidol (both CYP3A4 substrates) and morphine (not influenced by CYP3A4) had, as expected, no influence on fentanyl pharmacokinetics 33, 39.

In summary, most strong CYP3A4 inhibitors significantly increased systemic fentanyl exposure while moderate CYP3A4 inhibitors did not. All studied CYP3A4 inducers significantly decreased systemic fentanyl exposure. Although the effects were lower than expected, caution is warranted. Several case studies report interactions between strong CYP3A4 inhibitors and fentanyl with serious outcomes like respiratory depression or even death 40, 41, 42, 43. In daily clinical practice, use of strong CYP3A4 inhibitors may lead to higher fentanyl plasma levels and probably more (side) effects and use of CYP3A4 inducers may lead to lower fentanyl plasma levels and a risk of inefficient pain relief. Both interactions ask for careful monitoring of patients, especially when the interacting drug is started or stopped during treatment with fentanyl.

Environmental factors

No environmental factors have been studied with intravenous fentanyl.

Localization of the patch

Patients stick fentanyl patches at various sites on the body during chronic treatment. Preferred localizations are the upper arms, thorax and upper back. These different localizations may potentially influence fentanyl absorption as a result of differences in the thickness of the skin and the subcutaneous fat. Two studies analyzed whether the transdermal absorption from the patch differed between application sites by measuring the fentanyl residue in used patches. No differences in fentanyl absorption were found between patches applied to the arm, shoulder, chest and back 44, while the delivery efficiency of fentanyl was 7.5% lower for patches applied to the leg in comparison to the arm 39.

Local heat or cold

Heat, or raised temperature may increase locally the microcirculation and blood vessel permeability and thereby promote absorption of transdermally delivered drugs. Several case reports have been published about heat‐induced fentanyl toxicity (e.g sun bathing, warming blanket) in patients using a fentanyl patch 45, 46, 47. Three studies investigated the influence of heating the patch on the pharmacokinetics of fentanyl in a controlled setting. Despite some differences in study designs, study outcomes were similar. The AUC and maximum concentration (C _max) increased two‐ to fourfold when heat (40–44°C) was applied to the patch 28, 48, 49. In another study, the use of hot drinks before taking a dose of fentanyl spray sublingually had no effect on the absorption of fentanyl, neither did the use of cold drinks 50. Other rapid onset fentanyl products like buccal and sublingual tablets have not been studied yet in this setting. Theoretically, local heat may lead to local vasodilatation and probably a higher or faster (peak) concentration of fentanyl.

Other factors

Oromucosally delivered fentanyl is absorbed by transmucosal diffusion. In theory, intake of specific food or drinks before using these fentanyl products may influence fentanyl absorption and, therefore, pharmacokinetics. Experimental pre‐treatment with high pH beverages (solution of sodium bicarbonate in water) or low pH beverages (e.g. Coca‐Cola or Sprite) has been studied but did not influence absorption of fentanyl from a sublingual spray 50. The influence of both alcohol and smoking on the pharmacokinetics of fentanyl was studied in one study analyzing fentanyl residue in patches 44. No significant influence of both factors on fentanyl absorption rate was found. Finally, the effect of diurnal variation on fentanyl pharmacokinetics was studies during three consecutive days; no influence on its pharmacokinetics was found. Whether there is no effect at all of chronopharmacology on fentanyl pharmacokinetics needs to be shown in future studies 51.

In summary, only local heat applied to transdermal fentanyl patches significantly increased the AUC of fentanyl. In daily clinical practice, patients using fentanyl patches should be aware of the risk of a fentanyl intoxication in situations like fever, sun bathing, using a warming blanket and doing heavy exercise 45, 47, 52. All the other environmental factors studied did not significantly influence fentanyl pharmacokinetics.

Patient‐related factors

Various patients characteristics may potentially influence the pharmacokinetic IIV of fentanyl. All studied characteristics are summarized below, however not all characteristics are studied as extensively as others.

Age

Ageing is a multifactorial continuous process which may result in changed body composition, impaired renal function, decreased muscle tissue and comorbidity leading to changes in all different phases of pharmacokinetics. Age was studied widely as a potential factor influencing fentanyl pharmacokinetics 29, 37, 39, 44, 53, 54, 55, 56, 57, 58. The intravenous 29, 53, 56, transdermal 37, 39, 44, 54, 57, 58 and transmucosal 55 administration routes were studied. Even though all these studies investigated the relation between fentanyl pharmacokinetics and age, the methods used and pharmacokinetic parameters calculated were quite miscellaneous, making it difficult to draw solid conclusions.

Overall, five studies were explicitly designed to investigate pharmacokinetic differences between elderly and younger patients 29, 54, 55, 56, 58. Two of these studies used intravenous fentanyl, both performed in a perioperative setting 29, 56. In one of the perioperative studies, clearance was found to be significantly lower in older patients 29. Similar conflicting results were found in the three studies using other administration routes: in one of the two studies using transdermal fentanyl, a higher AUC was found for elderly patients 54, the study on transmucosal fentanyl found no effect of age 55.

Although lower fentanyl clearance in elderly patients may be explained by lower plasma albumin, decreased hepatic blood flow or decreased renal function, the exact reason remains unclear in the studies on intravenously administered fentanyl 59. In particular, confounding factors during surgery, such as the use of anaesthetics and effects on (portal) blood flow during abdominal surgery, may contribute to results in the perioperative studies 29, 56, 58.

In other studies age was one of several factors studied (in multivariate analyses) 37, 39, 44, 57. In all studies transdermal fentanyl was used. These studies did not show significant effects of age on various pharmacokinetic parameters, except for the study by Solassol 44. A small difference of 10% in mean fentanyl absorption (residue in patch) was found between patients >75 years compared to patients <65 years, with lower absorption rates for elderly patients. This variation is most likely too small to cause different clinical effects between elderly and younger patients in daily clinical practice.

In summary, although ageing may influence fentanyl pharmacokinetics, it is difficult to draw solid conclusions. There is at least a risk on lower clearance and thus higher AUCs in elderly patients. Therefore, our advice is to titrate fentanyl cautiously in elderly patients.

Gender

The effect of gender on fentanyl PK has been studied only in patients using transdermal patches. Gender may influence fentanyl pharmacokinetics by a higher CYP3A4 activity in women compared to men and by differences in body composition between men and women 60. Three studies investigated whether gender influences the pharmacokinetics of fentanyl patches; two analyzed residue in patches while one studied pharmacokinetics by serum samples 21, 39, 43. In these studies gender did not influence absorption from the patch 39, 43. Men had lower serum fentanyl concentrations than women but the influence of gender was quite small, as less than 1% of the inter‐individual variations was explained by gender 21. Gender may influence elimination, suggested by a higher average urinary fentanyl excretion for men than for women 39. It is questionable, however, how important this is because fentanyl is mostly metabolized in the liver, and only 10% is excreted unchanged in the urine. So, based on these studies, there is no reason for gender‐specific dose modifications of fentanyl.

Albumin

Fentanyl is highly lipophilic and binds to plasma proteins (e.g. albumin, alpha‐1‐acid glycoprotein). The effect of hypoalbuminemia was studied in two studies with a different design, both in patients using a fentanyl patch 21, 61.

In the study by Nomura et al., serum fentanyl concentrations were measured every 3 h during conversion from intravenous fentanyl to transdermal patch(es). After 6 h the intravenous fentanyl was stopped. The patients with an albumin <3.5 g dl⁻¹ had significantly reduced dose‐adjusted serum fentanyl concentrations at 9–24 h after application of the patch compared to patients with an albumin >3.5 g dl⁻¹. The authors suggest lower absorption from the patch in patients with low plasma albumin concentrations 61. Barrett et al. showed no clinically relevant influence of albumin on fentanyl plasma concentrations in a large cross‐sectional study in patients using transdermal fentanyl 21.

Neither study measured the free unbound fraction of fentanyl to better clarify the influence of plasma proteins on fentanyl pharmacokinetics. Also the cause of hypoalbuminaemia (cachexia, liver failure) may influence fentanyl pharmacokinetics. The cause of the hypoalbuminaemia may predict also if fentanyl pharmacokinetics are influenced in other ways than just absorption. A new study with intravenous fentanyl would be helpful to determine this. In daily practice, it is important to realize that higher doses of transdermal fentanyl may be needed to reach adequate pain relief when albumin levels decrease during phases of disease.

Body Mass Index (BMI)

The body mass index (BMI), or quetelet‐index (QI), is the most widely used index to indicate whether a person is under‐ or overweight. It is calculated by weight (in kg) divided by height in m². As BMI is related to body composition and the thickness of subcutaneous fat, transdermal fentanyl absorption and tissue distribution is expected to be larger in patients with high vs. low BMI. Several studies investigated the effect of BMI on fentanyl pharmacokinetics. Intravenous fentanyl was not studied, all studies included cancer patients using patches 21, 39, 44, 57, 61, 62. Only Heiskanen et al. found a statistically significant difference in plasma fentanyl concentration between cachectic patients and normal weight patients 62. At 48 and 72 h after applying the patch, plasma fentanyl concentrations were almost twice as low in cachectic patients compared to normal weight patients. A limiting factor of this study was that serum albumin concentrations were not measured in this study 62. Nomura et al. found no differences in absorption between low weight and normal weight patients during the first 24 h after conversion from intravenous to transdermally delivered fentanyl 61. In that study median serum albumin concentrations were similar in the low weight and normal weight patient groups. Thus, the lower plasma fentanyl concentrations found by Heiskanen et al. in the cachectic patients might be explained by hypoalbuminaemia 62.

Three studies analyzed the fentanyl residue in patches. None of these studies showed significant differences between different BMI groups 39, 44, 57. BMI is only studied in patch studies and not in intravenous and/or oromucosal administration routes. So, besides its possible influence on absorption, it is unclear what other parts of fentanyl metabolism it might also influence.

One study found significant lower fentanyl concentrations in extremely cachectic patients using fentanyl patches. Other studies did not find differences between low weight and normal weight patients. In daily practice physicians should be aware that fentanyl patches may be less effective in cachectic patients.

Renal function

Although fentanyl is mainly metabolized in the liver into the inactive metabolite norfentanyl, about 10% of both compounds are excreted by the kidneys. Three studies investigated whether kidney function influences fentanyl pharmacokinetics 21, 36, 39. Koehntop et al. studied intravenous fentanyl clearance in a specific setting, eight patients undergoing renal transplantation for terminal kidney dysfunction. They found that a blood urea nitrogen (BUN) > 60 mg dl⁻¹ was associated with a lower clearance compared to a BUN < 49 mg dl⁻¹. This effect probably reflected the heterogeneity in dialysis status, renal failure induced abnormalities and, probably most important, dynamic changes during surgery in this small study group 36. In a study in which the excretion of fentanyl by urine was measured in patients using fentanyl patches, elimination of fentanyl was not influenced by moderate to severe renal impairment 39. In this study, 20% of the patients had moderate or severe renal impairment defined as glomerular filtration rate (GFR) between 15 and 59 ml min⁻¹ 1.73m⁻². GFR and kidney disease were also part of the multivariate analysis of Barrett et al. 21. GFR was not associated with serum fentanyl concentrations. In daily practice there is no reason to adjust fentanyl dose depending on renal status.

Liver function

Fentanyl is mainly metabolized in the liver by CYP3A4 into inactive metabolites and, therefore, it is expected that liver disease will impair fentanyl clearance. Intravenous fentanyl is not studied; all studies report on transdermal fentanyl. In a study including patients with various degrees of liver failure, a clinically significant effect on pharmacokinetics was found. In the study by Kokubun et al., the Child Pugh Score was used to describe the severity of liver disease. Child Pugh score A is defined as mild (5–6 points), B as moderate (7–9 points) and C as severe (10–15 points) liver disease. The study showed that in severe liver failure, the AUC of fentanyl was increased; a higher Child Pugh Score was related to a lower clearance of fentanyl. Severe liver failure led to a seven times lower clearance of fentanyl compared to mild liver failure 37. The finding of clearly decreased urinary elimination of fentanyl in liver failure in a study measuring fentanyl excretion in urine was in line with the Kokubun study 39. Both studies showed that an impaired liver function influences fentanyl pharmacokinetics importantly. Therefore, fentanyl doses have to be adjusted when patients develop liver failure.

Mucositis

Mucositis is a painful inflammation and ulceration of the mucosa of the gastrointestinal tract. It may develop anywhere along this tract. Oral mucositis refers to mucositis of the mouth and occurs often in cancer patients during treatment with chemotherapy or radiation. Nowadays several rapid‐onset products of fentanyl are available 10. The sublingual and buccal rapid‐onset fentanyl products are absorbed by the oral mucosa followed by systemic delivery. Changes in the mucosal integrity due to mucositis have been hypothesized to influence the absorption of fentanyl. Two studies have been performed comparing the pharmacokinetics of bucally delivered fentanyl products in cancer patients with and without mucositis 63, 64. Both studies only included patients with a clinical grade 1 mucositis. Grade 1 mucositis consists of erythema, painless ulcers or mild soreness (CTCAE criteria). In neither study was a statistically significant difference in the pharmacokinetic parameters C _max, T _max and AUC found between the patients with and without mucositis, although one of the studies showed a trend towards a higher AUC (median 2.05 ng h⁻¹ ml⁻¹ vs. 1.55 ng h⁻¹ ml⁻¹) in patients with mucositis 63. Whether transmucosally fentanyl products can be safely and effectively used in regular doses in patients with more severe mucositis needs to be studied in adequately powered studies, including patients with more severe and, in that case, painful mucositis.

Rhinitis

Rhinitis is defined as irritation and inflammation of mucosa inside the nose caused by allergens, viruses, bacteria or irritants. In one study the effect of rhinitis and treatment with oxymetazoline was studied. In patients prone to develop allergic rhinitis, pharmacokinetics of intranasal fentanyl were measured in periods with or without rhinitis and with or without treatment with oxymetazoline. Rhinitis per se did not influence fentanyl pharmacokinetics, but the use of oxymetazoline reduced the C _max by almost 50%, most likely caused by vasoconstriction by oxymetazoline. Patients using intranasal fentanyl should not use concomitant local vasoconstrictives because of dramatically reduced concentrations of fentanyl probably leading to insufficient pain relief 65.

Hypertrichosis and hyperhidrosis

Hypertrichosis is an abnormal hair growth over the body and hyperhidrosis is a disorder marked by excessive sweating. Both factors are studied in a patch study analyzing the residue in patches. Both factors could potentially influence local adherence of the patch on the skin. However, neither factor influenced absorption in univariate analyses. Unfortunately, the severity of the hypertrichosis and hyperhidrosis – especially at the location of the fentanyl patch – were not described in the publication, making it difficult to interpret these findings 44.

Burns

Patients with severe burns start in a hypodynamic state immediately after the accident and end up in a hypermetabolic state represented by an increased cardiac output and reduced systemic vascular resistance 66. Due to these haemodynamic changes, pharmacokinetic characteristics may be influenced. One study investigated the influence of severe burns on the pharmacokinetics of intravenous fentanyl in patients scheduled for burn‐related surgery in their hyperdynamic phase. Patients with major burns (mean 49% ± 3% burn of body surface area) were compared to demographically matched controls. In the patients with burns, clearance of fentanyl was about 44% higher than in the control group. This higher clearance may be caused by increased cardiac output and the resultant increased hepatic blood flow 67. Although severe burns have significant impact on fentanyl clearance, this is not a common scenario for most cancer patients using fentanyl patches.

Pharmacogenetics

The effects of genetic variation on fentanyl pharmacokinetics have been studied in only a few trials so far 21, 38, 68, 69. Two studies used intravenous fentanyl 38, 69, the other studies used transdermal fentanyl. As mentioned earlier, fentanyl is thought to be mainly metabolized by CYP3A4. Of note, CYP3A5 contributes to CYP3A‐dependent drug clearance and thus may lead to changes in fentanyl pharmacokinetics as well 70.

Patients with the CYP3A5*3 gene single nucleotide polymorphism (SNP) had about a 2‐fold higher fentanyl plasma concentration normalized by measured absorption rate than patients with the wild‐type (*1*1) gene polymorphism and the patients with the heterozygous (*1*3) gene polymorphism 68, 69. The total clearance of fentanyl is also 30–50% lower for the *3*3 group compared to the other two groups 68, though Barrett et al. found no influence of CYP3A5*3 or the recently discovered CYP3A4*22 SNP on serum fentanyl concentrations 21. This discrepancy may (partly) be caused by the timing of taking blood samples. In the study by Barrett et al., one random sample was taken during fentanyl use, while the samples in the other two studies for all patients were exactly timed 68, 69. Next to enzymes, also the efflux drug transporter ABCB1 (P‐glycoprotein) was studied. This protein is responsible for the transport of fentanyl through the blood brain barrier (a.o.) 71. SLCO1B1 is another protein responsible for transport 38. Variations in genes coding for these proteins may therefore influence fentanyl pharmacokinetics. However both the ABCB1 1236 polymorphisms and SLCO1B1*1a and *15 polymorphisms were not found to influence fentanyl pharmacokinetics 38, 68. So, CYP3A5 polymorphisms may influence, the fentanyl PK. However, more research is needed before implementing genotyping in clinical practice.

Discussion and future perspectives

In this review an overview of currently studied factors in relation to the pharmacokinetics of fentanyl is provided. Awareness of these different factors that influence fentanyl pharmacokinetics is important to prevent over‐ and underdosing of fentanyl, leading to intoxication or insufficient pain relief. This is especially relevant in case of opioid rotation. The most pronounced effects on fentanyl PK can be expected when given in combination with strong CYP3A inhibitors, or inducers or in case of impaired liver function. In these cases, patients should be monitored closely, especially with changes in the prescription of (combinations of) strong CYP3A4 inhibitors and inducers, or in case of deteriorating liver function.

Another important factor leading to clinically relevant increases in fentanyl exposure, is the adding of local heat to a fentanyl patch. This directly promotes the absorption of fentanyl and should therefore be avoided 15, 16, 26, 27, 28, 33, 34, 37, 38.

Conflicting results were reported for the factors age, BMI and gender. This is particularly due to the enormous heterogeneity of the included populations (healthy volunteers, (peri‐)operative patients and cancer patients), the methods used in these studies with only a minority of the studies performed with intravenous fentanyl and the studied pharmacokinetic outcome parameters. Many studies did not report clinically important pharmacokinetic parameters like AUC, T _max and t _1/2. Furthermore, power analyses for prespecified pharmacokinetic endpoints were not described in the majority of studies. The studies to investigate the effect of BMI and gender were done in patients/volunteers using patches and in this way absorption may be the major cause of inter‐patient variability. A factor influencing clearance is in this case less easy to detect.

In general, the prevalence of cancer is especially high in elderly people. More knowledge on the effect of age in relation to fentanyl PK would therefore be helpful in adequate dosing of fentanyl. Other common patient variables like smoking habits and use of alcohol are also sparsely studied. These factors are prone to change during different phases of disease. Possibly, these factors explain partly the wide variety in fentanyl PK. Fentanyl is widely used for cancer‐related pain, as many patients and health care professionals prefer a patch for drug delivery for reasons of convenience. The patch is especially appropriate for specific patient populations like patients with swallowing disorders, bowel obstruction and patients at the end of life. However, typical problems in these specific patient populations are cachexia and dehydration. The influence of these factors on fentanyl uptake and clearance is still largely unclear. Cancer patients usually use several drugs for other diseases or intercurrent problems (e.g. infectious problems), but also medication to treat the side effects of cancer therapies. Most of these drugs are hardly studied for their effects on fentanyl clearance. Therefore, it is unclear whether commonly used co‐medication like clarithromycin, verapamil or aprepitant (particularly used by cancer patients to treat chemotherapy‐induced nausea and vomiting) influence fentanyl pharmacokinetics. Also (strong) CYP3A4 inducers are sparsely studied, e.g. for phenytoin or St. John's wort the influence is currently unclear, although these effects are probably clinically relevant.

Of note, most published studies were performed in patients using a fentanyl patch or intravenous fentanyl. Nowadays, several other fentanyl products are available; the rapid onset opioids (ROOs), for the treatment of breakthrough pain, for example. Only six studies in this review studied one of the ROOs 15, 50, 55, 63, 64, 65 and just two studies were performed in cancer patients 63, 64. Although the metabolism of fentanyl is the same for all fentanyl products, transmucosal absorption may be influenced by local factors like mucositis or dry mouth. The studies in this review that investigated mucositis included only patients with a low grade mucositis and no influence on fentanyl pharmacokinetics was found 63, 64. Unfortunately, patients with painful (higher grade) mucositis were not included. For this reason, at the Erasmus MC Cancer Institute we are currently performing a pharmacokinetic study with sublingual fentanyl in patients with at least grade 2 mucositis caused by radiotherapy in combination with cisplatin or cetuximab in head and neck cancer patients (www.trialregister.nl; study number NTR4995).

Furthermore, we found no studies on the effect of xerostomia on the absorption of fentanyl using ROOs for sublingual and buccal use. Although patients are advised to rinse their mouth with water before taking the drugs, evidence on a protective effect in dry mouth is not available in the literature. Since dry mouth is a common side effect, e.g. in patients using opioids and patients formerly treated for head and neck cancer, studies on the pharmacokinetics and clinical effects of ROOs in these patient groups are awaited.

A few trials included in this study showed that a part of the variation in fentanyl concentration can be explained by the CYP3A5*3 SNP. However, not all studies showed the same effect and the effects of the investigated polymorphisms in relation to fentanyl pharmacokinetics were small and do not support routine genotyping in clinical practice.

An important limitation of this study is that we only investigated pharmacokinetic variability of fentanyl. The investigated covariates were not correlated to pharmacodynamic effects in terms of side effects and pain relief. We assumed that changes of more than 25–30% lead to clinically relevant effects. However, a clear relation between fentanyl pharmacokinetics and the incidence and severity of fentanyl‐induced side effects has not yet been demonstrated.

We have chosen to describe the main characteristics of the included studies without using a specific tool to assess the quality of the selected studies (Table 1). In this review, we aimed to describe as many potential factors as possible that influence fentanyl pharmacokinetics independently of the kind of study or outcome. This could be a limitation of our search, but provides the most broad overview currently possible in the field of fentanyl pharmacokinetics.

In summary, in this review we found several factors influencing fentanyl pharmacokinetics, but we still cannot completely explain the wide intra‐ and inter‐patient variability 14, 21, 22, 23, 24, 72. During the next years, we hope and expect that new data will become available to further unravel the complex pharmacokinetics of fentanyl in both cancer‐ and non‐cancer‐related pain. In our view, prospective research on fentanyl pharmacokinetics should be more focused on cancer patients using various fentanyl products, during several phases of the disease trajectory (curable and non‐curable disease) and on the relation between pharmacokinetics and clinical effects, both pain relief and side effects.

Competing Interests

All authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: no support from any organization for the submitted work. RH has received a grant from NutsOhra during the conduct of the study. CvR has received a grant from Prostrakan for an investigator‐initiated study and fees from Ipsen and Vifor to the institute. There are no other relationships or activities that could appear to have influenced the submitted work.

Appendix 1.

Embase

(Fentanyl:de,ab,ti OR Phentan*:ab,ti OR Sublimaze:ab,ti OR Fentora:ab,ti OR ‘R 4263’:ab,ti OR R4263:ab,ti OR Duragesic:ab,ti OR Durogesic:ab,ti) AND (Pharmacokinetic*:de,ab,ti OR kinetic*:de,ab,ti OR ((absor*:de,ab,ti OR (biological NEXT/1 transport*):de,ab,ti OR (tissue NEXT/1 distribut*):de,ab,ti OR biotransform*:de,ab,ti OR elimin*:ab,ti OR toxic*:de,ab,ti) AND (dosage*:de,ab,ti OR dosis:de,ab,ti OR dosing:de,ab,ti OR doses:de,ab,ti OR dose:de,ab,ti OR metabol*:de,ab,ti ))) AND (transderm*:de,ab,ti OR (trans NEXT/1 derm*):de,ab,ti OR oral:de,ab,ti OR buccal:de,ab,ti OR subling*:de,ab,ti OR (sub NEXT/1 ling*):de,ab,ti OR nasal:de,ab,ti OR subcutan*:de,ab,ti OR intraven*:de,ab,ti OR (trans NEXT/1 muc*):de,ab,ti OR (sub NEXT/1 cutan*):de,ab,ti OR (intra NEXT/1 ven*):de,ab,ti OR transmuc*:de,ab,ti OR (method* NEAR/3 administ*):de,ab,ti OR ‘drug administration route’/exp) NOT ([animals]/lim NOT [humans]/lim) AND ([English]/lim OR 74/74lim) NOT ([child]/lim NOT [adult]/lim) NOT ([editorial]/lim OR [letter]/lim OR [review]/lim OR [conference abstract]/lim OR [conference paper]/lim OR [conference review]/lim).

PubMed

(Fentan*[tw] OR Phentan*[tiab] OR Sublimaze[tiab] OR Fentora[tiab] OR R‐4263[tiab] OR R4263[tiab] OR Duragesic[tiab] OR Durogesic[tiab]) AND (Pharmacokinetic*[tw] OR kinetic*[tw] OR ((absor*[tw] OR biological transport*[tiab] OR tissue distribut*[tw] OR biotransform*[tw] OR elimin*[tiab] OR toxic*[tw]) AND (dosage*[tiab] OR dosis[tiab] OR dosing[tiab] OR doses[tiab] OR dose[tiab] OR metabol*[tiab]))) AND (transderm*[tw] OR trans derm*[tw] OR oral[tw] OR buccal[tw] OR subling*[tw] OR sub ling*[tw] OR nasal[tw] OR subcutan*[tw] OR sub cutan*[tw] OR intraven*[tw] OR intra ven*[tw] OR transmuc*[tw] OR trans muc*[tw] OR methods of administ*[tw] OR administration method*[tw] OR Drug Administration Routes[mesh]) NOT (animals[mesh] NOT humans[mesh]) NOT (children[mesh] NOT adults[mesh]) AND (english[lang] OR Dutch[lang]) NOT (letter[pt] OR review[pt] OR editorial[pt]).

Cochrane

(Fentanyl:ab,ti OR Phentan*:ab,ti OR Sublimaze:ab,ti OR Fentora:ab,ti OR ‘R 4263’:ab,ti OR R4263:ab,ti OR Duragesic:ab,ti OR Durogesic:ab,ti) AND (Pharmacokinetic*:ab,ti OR kinetic*:ab,ti OR ((absor*:ab,ti OR (biological NEXT/1 transport*):ab,ti OR (tissue NEXT/1 distribut*):ab,ti OR biotransform*:ab,ti OR elimin*:ab,ti OR toxic*:ab,ti) AND (dosage*:ab,ti OR dosis:ab,ti OR dosing:ab,ti OR doses:ab,ti OR dose:ab,ti OR metabol*:ab,ti ))) AND (transderm*:ab,ti OR (trans NEXT/1 derm*):ab,ti OR oral:ab,ti OR buccal:ab,ti OR subling*:ab,ti OR (sub NEXT/1 ling*):ab,ti OR nasal:ab,ti OR subcutan*:ab,ti OR intraven*:ab,ti OR (trans NEXT/1 muc*):ab,ti OR (sub NEXT/1 cutan*):ab,ti OR (intra NEXT/1 ven*):ab,ti OR transmuc*:ab,ti OR (method* NEAR/3 administ*):ab,ti).

Kuip, E. J. M. , Zandvliet, M. L. , Koolen, S. L. W. , Mathijssen, R. H. J. , and van der Rijt, C. C. D. (2017) A review of factors explaining variability in fentanyl pharmacokinetics; focus on implications for cancer patients. Br J Clin Pharmacol, 83: 294–313. doi: 10.1111/bcp.13129.