Title: Adrenal insufficiency in patients on long-term opioid analgesia

Short title: Adrenal insufficiency and opioid analgesia

Authors:

Dr Fraser W Gibb1

Dr Alexandra Stewart2

Prof Brian R Walker1

Prof Mark W J Strachan1

Affiliation:

1. Edinburgh Centre for Endocrinology and Diabetes, NHS Lothian and

University of Edinburgh

2. Department of Anaesthetics, Royal Infirmary of Edinburgh

Correspondence:

Dr Fraser W Gibb

Edinburgh Centre for Endocrinology and Diabetes

Royal Infirmary of Edinburgh

Edinburgh

EH16 6AG

United Kingdom

Email: fraser.gibb@nhs.net

Keywords: Cortisol, adrenal, opioid, morphine

Disclosure: nothing to declare

Word count: 1933

Acknowledgements: This project was supported by a grant from the Edinburgh

and Lothians Health Foundation. We are grateful to the CORtisol NETwork

(CORNET) consortium for making available cortisol data from population-based

cohorts.

ABSTRACT

Objective:

Opioid analgesia has been implicated as a cause of secondary adrenal

insufficiency but little is known of the prevalence of this potentially serious

adverse effect in patients with chronic pain.

Design:

Cross-sectional study of chronic pain patients on long-term opioid analgesia.

Patients:

Patients attending tertiary chronic pain clinics at the Western General Hospital,

Edinburgh, treated with long-term opioid analgesia (n=48) with no recent

exposure to exogenous glucocorticoids.

Results:

4 patients (8.3%) had basal morning plasma cortisol concentrations below 100

nmol/L, of whom 3 failed to achieve a satisfactory cortisol response to

exogenous ACTH1-24 stimulation (peak cortisol >430 nmol/L). Basal cortisol was

positively associated with age (R = 0.398, p = 0.005) and negatively associated

with BMI (R = -0.435, p = 0.002).

Conclusions:

Suppression of the hypothalamic-pituitary-adrenal axis is present in a clinically

significant proportion of chronic pain patients treated with opioid analgesia.

Studies of larger populations should be conducted to better define the

prevalence and potential clinical consequences of adrenal insufficiency in this

context.

Introduction:

The use of long-term opioid analgesia is increasingly common, rising by

approximately 10% from 2003 until 2013 in England, when over 6.4 million

prescriptions were issued for morphine sulphate, buprenorphine and oxycodone

annually1. Compared with nonsteroidal anti-inflammatory drugs, opioid

analgesics may be associated with a higher risk of mortality in older individuals2,

although mechanisms underlying this association remain unclear. In recent

years a number of case reports have described cortisol deficiency in patients

treated with a variety of opioid analgesics, both oral3 and transdermal4. Opioids

affect the circadian rhythm of the hypothalamic-pituitary-adrenal (HPA) axis5

and reduce the effect of exogenous CRH upon ACTH and cortisol release6.

The suppressive effects of opioids upon the hypothalamic-pituitary-gonadal axis

have been well characterized, with hypogonadism present in 75% of male and

21% of female chronic pain patients treated with opioids7. However, the

prevalence of adrenal insufficiency has not been systematically assessed, except

in chronic intra-thecal opioid administration, where it is estimated to occur in

approximately 15% of patients8. We investigated the prevalence of cortisol

insufficiency in opioid-treated patients, attending a tertiary chronic pain clinic,

and whether this was associated with typical clinical manifestations of cortisol

insufficiency, such as hypotension and weight loss.

Methods:

Potential participants were identified from the electronic records of patients

attending chronic pain clinics at the Western General Hospital, Edinburgh.

Patients were eligible for inclusion if treated for at least 6 months with opioid

analgesics in the setting of chronic non-cancer pain. Patients with disease of the

HPA axis, current glucocorticoid therapy (by any route) or glucocorticoid

therapy within the preceding 3 months were excluded. 236 patients were

contacted by letter to participate in the study, with 48 ultimately taking part. Of

the 236 contacted, 8 patients were excluded from the study due to

current/recent glucocorticoid use and 4 were excluded due to recent

discontinuation of opioid medication. There were no significant differences in

age (54 [IQR 45 – 62] vs. 49 [42 – 59] years, p = 0.084) or gender representation

between those taking part and those who did not (17.6% of women and 24.5% of

invited men took part, p = 0.248). Ethical approval was granted by the South

East Scotland Ethics Committee and written informed consent obtained from all

participants.

Participants attended the clinical research facility at 8am and completed the SF-

36 health questionnaire. Basic anthropometric measurements were obtained

(weight and height) and body fat assessed using an Omron BF306 bioimpedance

device. Both erect and supine blood pressure and heart rate were measured

using an Omron 705IT automatic blood pressure monitor. Blood sampling was

performed immediately before and 30 min after IM administration of 250

micrograms of ACTH1-24 (Synacthen®). Cortisol was measured by Abbott

Architect Immunoassay, with a stimulated cortisol >430 nmol/L regarded as

normal, in keeping with locally established normative data. When patients

failed to achieve a sufficient cortisol response to ACTH stimulation, full anterior

pituitary function testing was performed in line with standard clinical care

including, if appropriate, pituitary MRI.

Normative data for morning cortisol were obtained from two European

population-based cohorts: CROATIA-Korcula (n=964) and ORCADES (n=1019) 9.

Cortisol was measured by immunoassay from samples taken between 0700h and

1100h. In addition, mean and standard deviation data were provided by Abbott

Laboratories (personal communication) from morning cortisol values (sampled

before 10 am) used to produce the Architect immunoassay reference range (n =

150). These data were used to create z-scores to permit direct comparison with

results obtained in the current study.

Dose of opioid is presented as a morphine sulphate (MST) dose equivalent as

described in the British National Formulary10. Correlations are presented as

Pearson correlation coefficients. Continuous variables were compared by

Kruskal-Wallis test. Data are presented as median (interquartile range). P value

<0.05 was regarded as significant for all analyses.

Results:

Demographics and opioid treatment

The median age of participants was 53.5 years (45.4 – 62.4). 25 participants

were female and 23 were male. Median BMI in the cohort was 31.2 kg/m2 (26.4

– 37.4). The median daily opioid dose was 68 mg (40 – 153) as determined by

morphine sulphate equivalent dose. A variety of opioid medications were

prescribed, as detailed in figure 1, with tramadol and dihydrocodeine treated

patients receiving significantly lower MST-equivalent doses.

Plasma cortisol

Cortisol at baseline and 30 min after ACTH1-24 stimulation is shown in figure 2.

8am cortisol was less than 100 nmol/L in 4 patients (8.3%). This represents

values below a z-score for the Abbott assay of -1.6. For comparison, only 1.1% of

unselected participants from the combined ORCADES / CROATIA-Korcula

population-based cohorts (n=1983) had morning cortisol values below the

equivalent threshold (z-score of -1.6). Three patients failed to achieve an

appropriate cortisol response to ACTH stimulation; all were female, aged

between 36 - 39 years, obese (BMI > 30mg/kg2), and with basal plasma cortisol

<100 nmol/L. When limiting this analysis to the 33 patients on high-dose opioid

analgesia (excluding tramadol and dihydrocodeine), approximately 10% of those

assessed had an initial suboptimal cortisol response to ACTH.

Clinical characteristics of the cohort are presented in table 1 and quality of life

questionnaire data in table 2, stratified by tertiles of basal cortisol. Lower basal

cortisol was associated with younger age (R = -0.435, p = 0.002) and higher BMI

(R = -0.435, p = 0.002) (figure 3), but not with other indices of glucocorticoid

action such as blood pressure and not with opioid dose. Lower basal cortisol

was associated with somewhat poorer quality of life scores (table 2). No

significant associations were observed between these variables and ACTH-

stimulated cortisol concentration (not shown).

Further investigation in patients with glucocorticoid insufficiency

Patient 1: A 39-year-old woman (BMI 37.9 kg/m2) treated with a fentanyl patch

(75 gμ /hour) and oramorph (5mg as required) [MST daily equivalent

185mg/day] for fibromyalgia. She had no additional co-morbidities. Initial

synacthen test showed a basal morning cortisol of 84 nmol/L rising to 270

nmol/L. Repeat synacthen tests showed borderline results 69 to 431 nmol/L

and 147 to 412 nmol/L. Morning ACTH was 11 ng/L (reference range 7 – 63

ng/L) and prolactin and IGF-1 were within the reference ranges. Mild subclinical

primary hypothyroidism was noted with a TSH of 13.5 mU/L and a free T4 of 11

pmol/L. She was receiving oral combined estrogen/progesterone replacement

therapy. Pituitary MRI was normal. She reported modest symptomatic

improvement following commencement of hydrocortisone (10mg am and 5 mg

pm), specifically with improvement in lethargy.

Patient 2: A 37-year-old woman (BMI 31.5 kg/m2) treated with MST (60mg

daily) and sevredol (40mg as required) [MST equivalent 100mg/day] for

neuropathic scar pain. She had no additional co-morbidities. Initial synacthen

test showed a basal morning cortisol of 91 nmol/L rising to 385 nmol/L. Repeat

synacthen tests showed borderline results 180 to 437 nmol/L and 257 to 375

nmol/L. Morning ACTH was 13 ng/L and prolactin, IGF-1 and thyroid function

were within the reference ranges. She had been amenorrhoeic for approximately

one year, following commencement of a depot contraceptive. Pituitary MRI was

normal. She reported symptomatic improvement following commencement of

hydrocortisone (10mg am and 5 mg pm), specifically with improvement in

fatigue, although this effect waned after approximately 6 months.

Patient 3: A 36-year-old woman (BMI 37.7 kg/m2) treated with a fentanyl patch

(75 gμ /hour) and dihydrocodeine (240mg daily) [MST daily equivalent

222mg/day] for chronic abdominal pain. She had no additional co-morbidities.

Initial synacthen test showed a basal morning cortisol of 64 nmol/L rising to 424

nmol/L. A repeat synacthen test was unequivocally normal (446 rising to 590

nmol/L) with an associated early morning ACTH of 37 ng/L. Full anterior

pituitary function testing was also normal and, in view of this, no further

investigations or treatment were considered.

Discussion:

In this series of opioid-treated chronic pain clinic patients, low morning cortisol

was more common than would be expected with reference to large population-

based studies. Similarly, 3 of 33 patients receiving high dose opioid failed to

mount an appropriate cortisol response to ACTH stimulation. In view of the

widespread use of opioid analgesia in chronic pain, a large number of patients

may be at risk of hypocortisolaemia.

Case reports3,4 and small case series11 have reported significant effects of opioid

analgesia upon the hypothalamic-pituitary-adrenal axis, although the prevalence

of adrenal insufficiency in this context has not been documented previously.

Prospective evaluation of plasma cortisol concentration in a series of 8 chronic

pain patients, following commencement of oral opioids, demonstrated significant

(possibly subnormal) reductions in morning cortisol, although ACTH-stimulated

cortisol was normal in all those investigated11. The circadian rhythm of ACTH

secretion is blunted in heroin addicts with lower morning concentrations

observed than in healthy controls12 and concomitantly lower morning plasma

cortisol, with attenuation of the expected late evening nadir13. Perhaps contrary

to expectations, opioid treated chronic pain patients were found to have elevated

hair cortisol levels when compared to healthy controls14. Elevated hair cortisol, a

measure of long-term systemic cortisol exposure, may however reflect changes

related to chronic pain rather than opioid therapy. Chronic pain has also been

associated with disruption of the diurnal rhythm of cortisol15, low morning

cortisol and enhanced reactivity of the HPA axis to stimulation16. Opioids may

have direct effects upon adrenal glucocorticoid production, based on previous

studies in rodents. In rat adrenal cell culture experiments, administration of

opioid peptides resulted in corticosterone production, which was diminished by

naloxone17. In vivo, morphine potentiated and naloxone inhibited corticosterone

production in hypophysectomised rats18. If similar mechanisms exist in humans,

a pattern of low morning cortisol, blunted ACTH-stimulated cortisol and loss of

diurnal rhythm could occur during chronic opioid exposure.

Demonstrably lower quality of life has been reported in chronic pain patients19

and cortisol insufficiency may compound this in opioid-treated patients, perhaps

mediated at a CNS level or in relation to attenuated anti-inflammatory

mechanisms. We were, however, unable to detect any significant association

between basal or stimulated cortisol and quality of life measures. The observed

relationship between basal cortisol and role limitation secondary to emotional

problems is of questionable significance. However, given the relatively small size

of this cohort, our findings do not necessarily exclude such relationships. In

older individuals, opioid analgesia has been associated with a higher mortality

risk than in matched patients treated with NSAIDs2. The mechanisms through

which opioids may exert such an effect remain obscure, although it would be of

significant interest to investigate the potential role of perturbations in the

hypothalamic-pituitary axes.

It is open to question whether biochemically modest cortisol deficiency is

associated with clinically meaningful symptoms or adverse outcomes, but it is

conceivable that during acute illness such patients may be at risk. When patients

with mild opioid-related adrenal insufficiency are identified, it is not clear

whether they benefit from glucocorticoid replacement. The current study was

not designed to address this question, although the two patients commenced on

hydrocortisone did report subjective improvement in their level of well-being.

The small size of the study limits the assessment of risk factors for

adrenocortical insufficiency and is inadequately powered to confirm dose-

response or drug-specific relationships with opioid exposure. Failure to mount

an appropriate cortisol response was limited to female patients who were

younger and with higher BMI than the cohort average. Sexual dimorphism has

been demonstrated in the propensity to develop opioid-induced hypogonadism7.

Further work is required to better establish the prevalence of cortisol deficiency

in patients receiving high dose long-term opioid therapy, as well as the clinical

relevance of modestly suboptimal cortisol response to ACTH-stimulation in this

context. Nonetheless, on the basis of the data presented in this study, we

suggest it reasonable to consider measuring an early morning plasma cortisol in

patients receiving long-term, high-dose opioids, with consideration of further

endocrine evaluation in those with a result less than 100 nmol/L.

References:

1. Health and Social Care Information Centre (2013) Prescriptions

Dispensed in the Community, Statistics for England – 2002 – 2012.

http://www.hscic.gov.uk/catalogue/PUB11291/pres-disp-com-eng-

2002-12-rep.pdf Accessed 3rd July 2015

2. Solomon, D.H., Rassen J.A., Glynn R.J. (2010) The comparative safety of

analgesics in older adults with arthritis. Arch Int Med, 170(22), 1968 –

76.

3. Müssig K., Knaus-Dittmann D., Schmidt H. et al. (2007) Secondary adrenal

failure and secondary amenorrhoea following hydromorphone treatment.

Clin Endocrinol, 66(4), 604-605.

4. Oltmanns K.M., Fehm H.L., Peters A. (2005) Chronic fentanyl application

induces adrenocortical insufficiency. J Intern Med, 257(5), 478 – 480.

5. Facchinetti F., Volpe A., Farci G. et al. (1985) Hypothalamus-pituitary-

adrenal axis of heroin addicts. Drug and alcohol dependence, 15(4), 361 –

366.

6. Rittmaster R.S., Cutler G.B., Sobel D.O. et al. (1985) Morphine inhibits the

pituitary-adrenal response to ovine corticotropin-releasing hormone in

normal subjects. J Clin Endocrinol Metab, 60(5), 891 – 895.

7. Fraser L.A., Morrison D., Morley-Forster P. et al. (2008) Oral Opioids for

Chronic Non-cancer Pain: Higher Prevalence of Hypogonadism in Men

than in Women. Exp Clin Endocrinol Diabetes, 117(01), 38 – 43.

8. Abs R., Verhelst J., Maeyaert J. et al. (2000) Endocrine consequences of

long-term intrathecal administration of opioids. J Clin Endocrinol Metab,

85(6), 2215 – 2222.

9. Bolton J.L., Hayward C., Direk N. et al. (2014) Genome wide association

identifies common variants at the SERPINA6/SERPINA1 locus influencing

plasma cortisol and corticosteroid binding globulin. PLoS genetics,10(7).

10. Joint Formulary Committee. British National Formulary (online) London:

BMJ Group and Pharmaceutical Press

<http://www.medicinescomplete.com> [Accessed on 30th June 2015]

11. Palm S., Moenig H., Maier C. (1997) Effects of oral treatment with

sustained release morphine tablets on hypothalamic-pituitary-adrenal

axis. Method Find Exp Clin Pharmacol. 19(4), 269 – 273.

12. Facchinetti F., Grasso A., Petraglia F. et al. (1984) Impaired circadian

rhythmicity of -lipotrophin, -endorphin and ACTH in heroin addicts. β β

Acta Endocrinologica, 105(2), 149-155.

13. Facchinetti F., Volpe A., Farci G. et al. (1985) Hypothalamus-pituitary-

adrenal axis of heroin addicts. Drug Alcohol Depend, 15(4), 361 – 366.

14. Van Uum S.H., Sauvé B., Fraser L.A. et al. (2008) Elevated content of

cortisol in hair of patients with severe chronic pain: a novel biomarker for

stress. Stress, 11(6), 483 – 488.

15. McCain G.A., Tilbe K.S. (1989) Diurnal hormone variation in fibromyalgia

syndrome: a comparison with rheumatoid arthritis. J Rheumatol Suppl,

19, 154 – 157.

16. Griep E.N., Boersma J.W., Lentjes E.G. (1998) Function of the

hypothalamic-pituitary-adrenal axis in patients with fibromyalgia and low

back pain. J Rheumatol, 25(7), 1374-1381.

17. Kapas S., Purbrick A., Hinson J.P. (1995) Action of opioid peptides on the

rat adrenal cortex: stimulation of steroid secretion through a specific mu

opioid receptor. J Endocrinol, 144(3), 503 – 510.

18. Heybach JP, Vernikos J. (1981) Naloxone inhibits and morphine

potentiates the adrenal steroidogenic response to ACTH. Eur J Pharmacol,

75(1), 1-6.

19. Becker N., Bondegaard Thomsen A., Olsen A.K. et al. (1997) Pain

epidemiology and health related quality of life in chronic non-malignant

pain patients referred to a Danish multidisciplinary pain center. Pain,

73(3), 393 – 400.

Figures

Figure 1: Regular opioid medication and median dose of opioid within each group (presented as equivalent daily dose of morphine sulphate [mg]).

Figure 2: Synacthen test results in study participants (n=48). Pearson R = 0.677 (p <0.001). Dotted line represent threshold for satisfactory cortisol response to ACTH-stimulation (>430 nmol/L).

Figure 3: Relationship of basal cortisol with (A) age (Pearson R = 0.398, p = 0.005) and (B) BMI (Pearson R = -0.435, p = 0.002).

Table 1: Clinical characteristics of patients stratified by tertiles of basal cortisol. Presented as median (IQR) and compared by Kruskal-Wallis test.

Morning cortisol tertilesTertile 1 Tertile 2 Tertile 3 p

Morning cortisol (nmol/L)

173 (106 – 237)

318 (271 - 343)

438 (384 - 466)

Stimulated cortisol (nmol/L)

490 (443 - 544)

596 (526 – 642)

619 (594 - 638)

Cortisol increment (nmol/L)

321 (273 - 372)

289 (241 - 332)

169 (137 - 225)

Age (years) 50.7 (40.5 - 54)

57.8 (48.2 – 64.8)

61.1 (49.5 – 68.2)

0.017

Gender F 9 / M 7 F 8 / M 8 F 8 / M 8Weight (kg) 90.9 (80.8 –

107.5)101.9 (87.5 – 115.3)

72.3 (67.7 – 91.8)

0.007

BMI (kg/m2) 32.4 (29.5 – 36.7)

31.5 (29.5 – 39.2)

26.0 (23.3 – 32.6)

0.008

Body fat (%) 39.2 (31.0 – 40.6)

32.3 (30.3 – 40.3)

33.6 (26.2 – 39.2)

NS

SBP supine (mmHg) 130 (120 - 136)

136 (125 - 158)

140 (126 - 157)

NS

DBP supine (mmHg) 75 (67 - 81) 84 (76 - 89) 78 (74 - 90) NSSBP erect (mmHg) 124 (117 -

132)135 (121 - 146)

136 (126 - 144)

NS

DBP erect (mmHg) 83 (73 - 86) 88 (76 - 104) 81 (77 – 89) NS SBP (mmHg)Δ 4 (-2 - 8) 7 (-3 - 12) 5 (-6 - 13) NS DBP (mmHg)Δ -8 (-11 - -2) -6 (-11 - -2) -3 (-8 - -2) NS

MST dose equivalent (mg/day)

95 (40 - 166) 75 (30 - 124) 60 (42 - 140) NS

Table 2: Results of SF-36 questionnaire stratified by tertiles of basal cortisol and presented as median (IQR).

Morning cortisol tertilesTertile 1 Tertile 2 Tertile 3 p

Physical functioning 20 (18 – 35) 15 (10 – 30) 20 (13 - 63) NSRole limitations due 0 (0 – 13) 0 (0 – 0) 0 (0 – 38) NS

to physical healthRole limitation due to emotional problems

0 (0 – 50) 0 (0 – 33) 33 (33 – 100) 0.047

Energy / fatigue 30 (15 – 43) 15 (8 – 25) 25 (15 – 33) NSEmotional well-being

56 (32 – 66) 52 (42 – 54) 68 (48 – 78) NS

Social functioning 19 (13 – 56) 25 (0 – 56) 38 (25 – 56) NSPain 23 (10 – 23) 23 (10 – 23) 23 (10 – 27) NSGeneral health 28 (15 – 45) 30 (15 – 33) 40 (20 – 45) NS