Drug‐induced Raynaud's phenomenon: beyond β‐adrenoceptor blockers

Abstract

Aim

Drug‐induced Raynaud's phenomenon (RP) has long been associated with the use of different drugs, including cancer chemotherapy or β‐adrenoceptor blockers. However, sources report extremely variable prevalence and the level of evidence for each class is heterogeneous. Moreover, new signals are emerging from case reports and small series. Our objective was therefore to review available evidence about this adverse drug effect and to propose a mechanistic approach of drug‐induced RP.

Methods

A systematic review of English and French language articles was performed through Medline (1946–2015) and Embase (1974–2015). Further relevant papers were identified from the reference lists of retrieved articles.

Results

We identified 12 classes of drugs responsible for RP, with a variety of underlying mechanisms such as increased sympathetic activation, endothelial dysfunction, neurotoxicity or decreased red blood cell deformability. Cisplatin and bleomycin were associated with the highest risk, followed by β‐adrenoceptor blockers. Recent data suggest a possible involvement of tyrosine kinase inhibitors (TKI), through an unknown mechanism.

Conclusion

Drug‐induced RP is a probably underestimated adverse drug event, with limited available evidence regarding its prevalence. Although rare, serious complications like critical digital ischaemia have been reported. When these treatments are started in patients with a history of RP, careful monitoring must be made and, if possible, alternative therapies that do not alter peripheral blood flow should be considered.

Introduction

Raynaud's phenomenon (RP) is characterized by transient ischaemia of the extremities in response to environmental stress or emotions 1. It typically manifests as changes to the fingers, with pallor (vasospasm and decreased blood flow), cyanosis (deoxygenation of the static venous blood) and rubor (reperfusion), often accompanied by pain. RP can be primary (i.e. idiopathic) or secondary to an underlying cause. In both cases, abnormalities of the cutaneous microcirculation are primarily involved in the pathophysiology of RP 2.

The prevalence of RP in the general population varies between 0.5 and 19%, with major geographic variability 3, 4, 5, 6. While primary RP is the most frequent form (80–90%) 7, RP may also be secondary to various auto‐immune diseases (such as systemic sclerosis (SSc), systemic lupus erythematosus, vasculitis, etc.), or other systemic diseases 1. Several drugs with peripheral vascular effects leading to decreased microvascular perfusion may induce or aggravate RP. Drug‐induced RP probably goes unrecognized because of the limited knowledge of this side effect.

Literature reviews and textbooks usually have comprehensively reviewed drugs that have long been known to be responsible for RP 8. However, new signals are emerging from numerous case reports. Yet, to our knowledge, no systematic review has been performed and little is known about the prevalence and the level of evidence of drug‐induced RP. Our objective in the present work was therefore to summarize available evidence and to propose a mechanistic approach of drug‐induced RP.

Methods

The MEDLINE database was searched for English or French language articles published between January 1946 and May 2015 using the following search terms: ‘Raynaud disease/chemically induced’ [MESH] and ‘raynaud’ AND ‘clonidine’, ‘betablocker’, ‘ergot alkaloid’, ‘dopaminergic agonist’, ‘selective serotonin re‐uptake inhibitors’, ‘sympathomimetic drugs’, ‘chemotherapy’, ‘tyrosine kinase inhibitors’, ‘interferon’ and ‘ciclosporin’. Further relevant papers were identified from the reference lists of retrieved articles. We used the Oxford Centre for Evidence‐based Medicine ‐ Levels of Evidence to graduate the strength of the link between RP and drug classes 9. Among 253 records screened, 131 full texts were assessed for eligibility and included is the review (Figure 1).

Figure 1.

Flow diagram of studies included in the review

Results

Drugs enhancing vasoconstriction

β‐adrenoceptor blockers

β‐adrenoceptor blockers have long been known as causing drug‐induced RP, but data about its prevalence are scarce. Analysis of the Framingham Heart study data identified β‐adrenoceptor blocker use as the most common cause of secondary RP (34.2% of secondary RP). A meta‐analysis published in 2012 that included 13 studies (1012 patients) found a prevalence of 14.7% in patients receiving β‐adrenoceptor blockers 4. However, the studies were old (1971 to 1984) and of varying quality. A network meta‐analysis of prospective randomized controlled trials revealed a prevalence of peripheral vasoconstriction among patients treated with β‐adrenoceptor blockers of 7% (1966/28072), whereas 4.6% (555/12060) and 1.7% (305/17492) of patients treated with placebo or active control experienced this adverse effect, respectively (P<0.001) (Khouri et al., submitted).

The pathophysiology of this side effect remains unclear. Studies exploring the effect of β‐adrenoceptor blockers on patients with primary RP failed to show any worsening of their symptoms 10, 11, 12, 13. There is no evident explanation for this discrepancy, but the studies have small sample size.

The influence of the ancillary properties of β‐adrenoceptor blockers (e.g. intrinsic sympathomimetic activity, β₁‐selectivity, vasodilator activity) should theoretically influence their propensity to induce peripheral vasoconstriction, although studies report conflicting results 13, 14, 15, 16. The recent network meta‐analysis conducted by our group suggests that β‐adrenoceptor blockers are a heterogeneous class. High affinity for β₁‐adrenoceptors does not protect from RP while ancillary properties such as intrinsic sympathomimetic activity and vasodilator properties seem to be protective (Khouri et al., submitted to British Journal of Clinical Pharmacology).

Clonidine

RP induced by clonidine is a well‐known adverse reaction, described since many years although its frequency is not known 17. In patients with RP, cold‐amplified α_2c‐adrenoceptors mediated vasoconstriction is increased 18, 19. It has been identified that skin vasoconstriction in response to local cooling is mediated by the translocation of α_2c‐adrenoceptors to the vascular smooth muscle cells surface, through a pathway involving RhoA–Rho kinase 20. In cold situations, clonidine direct α_2c‐vascular agonism may become pre‐eminent on the usually desired central reduction of the adrenergic tone.

Ergot alkaloids

Ergotamine and its derivatives are used to treat migraine disorders and cluster headache 21. They display affinity for a wide variety of receptors including those for 5‐HT (serotonin), dopamine and norepinephrine 22. They are partial agonists of various serotoninergic receptors and the usual response of blood vessels to 5‐HT is contraction 23. More precisely, they exert a central vasoconstrictor effect through serotoninergic 5‐HT_1B/1D receptors, which are mostly in the cranial vessels and at therapeutic dose exert only a weak constricting effect on peripheral blood vessels 24. However 5‐HT₂ agonism seems to be the main effector of their peripheral serotoninergic vasoconstrictor effect. Moreover, they are α₁‐_, α₂‐adrenergic and dopaminergic D₂‐receptor agonists. Numerous case reports illustrating this effect are found in the literature 25, 26. However, the accountability of ergot alkaloids in RP is difficult to assess because of a significantly higher prevalence of RP in the migraine population 27, 28. Furthermore, the peripheral vasoconstriction caused by ergot alkaloids is sometimes interpreted as RP. ‘Ergotism’ is rarely observed (estimated incidence is 0.1%), but the prolonged vasoconstriction can lead to gangrene.

In contrast, other drugs targeting serotonin receptors such as triptans, selective agonists of 5‐HT_1B/1D, do not induce vasoconstriction of extremities and RP.

Dopaminergic agonists

RP cases have been reported following the use of bromocriptine, another ergot alkaloid 29, 30, 31, 32. One report describes severe RP with vascular morphological injury (presence of megacapillary on nailfold capillaroscopy) attributed to 6 years of treatment with bromocriptine 31. Bromocriptine is mainly a dopaminergic agonist. At low doses it has vasodilatative properties resulting from D₁‐receptor activation and leading to the well‐identified orthostatic hypotensive state. At high doses it exhibits α₁‐adrenoceptor properties 33 and peripheral release of catecholamines both resulting in vasoconstriction. Moreover, direct activation of α₂‐adrenoceptors by bromocriptine has been described and could explain increased sensitivity to cold 34, like clonidine. Microvascular injury with long term use of bromocriptine has also been suspected. 31 Nevertheless, a large case–control study (542 cases and 2155 controls) did not support the association between dopamine agonists and an increased risk of ischaemic events requiring hospitalization 35. Unfortunately this study did not provide detailed information on RP.

Surprisingly, two cases of erythromelalgia have been described with bromocriptine, in association with calcium channel blockers 36, 37

Selective serotonin re‐uptake inhibitors (SSRIs)

Contradictory effects of SSRIs on peripheral vasoreactivity have been reported. On the one hand, SSRIs have been proposed as a treatment for RP, following the observation of the relief of patients with erythromelalgia or RP with fluoxetine and sertraline 38, or paroxetine and escitalopram 39. Indeed, fluoxetine blocks the uptake of serotonin by platelets and decreases the amount of serotonin that is released during platelet activation/aggregation, which may explain the favourable outcome in patients with primary or secondary RP participating in an open randomized clinical trial 40. On the other hand, other authors have described a deleterious association between RP and the SSRIs, fluoxetine 41, 42, fluvoxamine 43, citalopram 44 and milnacipran 45, together with the relief of erythromelalgia symptoms 46. A case of emerging RP 2 days after beginning tergaserod treatment, a partial 5‐HT₄ serotonin receptor agonist, has also been described 47.

Currently this discrepancy between vasoconstriction and vasodilatation remains unexplained. Some authors suggested that endothelial damage is necessary for the development of a vasoconstrictive effect during SSRI treatment 48. In a healthy vascular bed it has been proposed that blocking serotonin re‐uptake could increase free plasma serotonin concentrations and produce, in stasis conditions, a local accumulation of serotonin, exacerbating vasoconstriction through 5‐HT₂ receptors that may worsen RP 44. In contrast, SSRIs decrease the amount of serotonin that is released during platelet activation/aggregation. For example fluoxetine is known to deplete platelet serotonin by 95% 49. Individual variability in metabolism or in signalling serotonin pathways could explain this variability in response to SSRIs 40.

Stimulants

Central stimulation of the dopaminergic and noradrenergic system is responsible for the peripheral release of catecholamines leading to vasoconstriction. Cases of RP induced by central nervous system stimulants have been reported 50. A retrospective case–control study investigated whether medications used for the treatment of attention deficit hyperactivity disorder (ADHD) were associated with the development of RP. Sixty‐four children were enrolled in the study (32 cases with RP and 32 age and gender matched control patients) and a significant association between the presence of RP and past or current use of ADHD stimulants (methylphenidate and dextroamphetamine) was found 51. Atomoxetin, a selective norepinephrine re‐uptake inhibitor, was excluded from this study because it was not considered as a central nervous system stimulant. However the case of dose dependent RP following the use of atomoxetin on a girl has recently been described 52. Two cases of RP induced by reboxetin, an inhibitor of norepinephrine re‐uptake, have been described 47. Amphetamine‐like drugs have also been associated with the emergence of RP and vasculopathy, as has phentermine, a weak sympathomimetic agent, used most commonly as an appetite suppressant in the treatment of obesity 53.

Ciclosporin

A study assessing the prevalence of RP in 100 renal transplant patients treated with ciclosporin monotherapy who were then transferred to prednisolone and azathioprine observed the development of de novo symptoms in 39% of patients on the introduction of ciclosporin. After withdrawal of ciclosporin, symptoms improved in 89% 54. Moreover four case studies of RP induced by ciclosporin use have been described. Three of them appeared a few days after ciclosporin introduction and totally disappeared after cessation 55, 56. The second case was dose related but persistent RP symptoms were observed after cessation of ciclosporin 57. The mechanism for ciclosporin induced RP remains unclear. A vasospastic effect of ciclosporin on both the macro and microcirculation has been shown 58 leading to systematic monitoring of hypertension or acute renal failure in the early treatment phase. Furthermore, changes in the viscosity of the blood, a decrease in the deformability of red blood cells and an increase in the aggregation of platelets can also be induced by ciclosporin use and contribute to RP 55.

It is worth noting that drugs increasing blood viscosity such as erythropoietins or intravenous immunoglobulins are not a known cause of RP. Much about the physiopathology of drug‐induced RP remains to be learnt.

Sympathomimetics

Digital necrosis was described following the localized use of lidocaine/epinephrine in a patient with primary RP 59. Data concerning sympathomimetic nasal decongestants (pseudoephedrine, phenylephrine) are scarce. Thus, pharmacologic properties of these drugs and their poor clinical benefit suggest that they should be contraindicated in patients with scleroderma‐related RP 60.

Toxic substances

Among recreational drugs, RP with ischaemic finger necrosis was attributed to cocaine abuse in the case of in a 37‐year‐old man 61. Cocaine has a potent vasoconstrictor effect through its α₂‐adrenoceptor activity. In animal studies it has also been shown to alter prostaglandin production with disproportionate increases in thromboxane in rabbit endothelium resulting in vasospasm, platelet activation and thrombus formation 62, 63, although in some cases cocaine vasculopathy is more likely to be related to a Buerger‐like syndrome 64, 65, 66 as described with cannabis use or arsenic exposure 67, 68, 69.

Endothelium damage and/or neurotoxicity

Cancer chemotherapies

The link between RP and chemotherapies has long been clearly identified. First descriptions of chemotherapy‐induced RP were related to treatments for testicular cancer 70, 71, 72. A study in 1995 that included 90 patients treated with cisplatin‐based chemotherapy for more than 1 year after testicular cancer found that 37% of them had developed RP after four cycles of chemotherapy combining cisplatin, bleomycin and vinblastine 73. RP typically appeared 3 to 6 months after the start of chemotherapy and often persisted for several years 74. The risk factors identified for the development of RP were high cumulative doses of bleomycin and a combination of bleomycin with vinblastine rather than etoposide.

Furthermore, a trend towards the increased prevalence of RP was observed in patients who received bleomycin as a bolus compared with continuous infusion. No significant correlation was seen with the cumulative or single doses of cisplatin, etoposide or vinblastine, serum magnesium concentrations during or after chemotherapy or a history of smoking 73.

These results were confirmed by the follow‐up of a cohort study that included 739 patients treated for testicular cancer between 1982 and 1992. Patients were divided between chemotherapy (n = 384) and non‐chemotherapy (n = 355) groups. The prevalence of RP was significantly higher among patients who received chemotherapy (20.7% vs. 1.7%, P<0.001) 75. Once again, a significant relationship between the cumulative dose of bleomycin and the prevalence of RP was found (OR 2.98, 95% CI, 2.286, 3.388, P<0.001); P<0.001). Thirteen percent of patients still suffered from RP 10 years after having received a cumulative bleomycin dose of <180000 IU (corresponding approximately to three cycles of cisplatin‐etoposide‐bleomycin), 24.6% after a cumulative dose of 180 000 IU to 360 000 IU, and 29% after a cumulative dose >360 000 IU. A large observational study 76 including 1409 testicular cancer survivors found a prevalence of RP among the chemotherapy group of 39%. The cancer chemotherapy associated Vinca alkaloids, cisplatin and bleomycin. The odds ratios for Raynaud‐like phenomena in those who received one to four cycles of chemotherapy compared with those who received no chemotherapy were 2.9 [95% CI, 2.2, 3.9] and 8.0 [95% CI, 4.4, 14.7] if they received more than five cycles of chemotherapy. When these drugs have been used to treat Kaposi's sarcoma RP has also been described 77, 78, 79, 80, 81. Nevertheless, emergence of severe RP with digital necrosis after a single cycle of doxorubicin, bleomycin, vincristine and dacarbazine chemotherapy, with a cumulative dose of only 40 000 IU of bleomycin has also been described 82. Cases describing the occurrence of RP after the local injection of bleomycin to treat warts have also been reported 83, 84, 85, 86, 87.

While RP has been associated with cisplatin‐based chemotherapies the immutability of cisplatin itself remains unclear 73, 88. A recent meta‐analysis of cisplatin‐based chemotherapies included 24 studies (n = 2479 patients) and found a prevalence of RP of 24% (95% CI, 17.5, 31.3) 72. However, cisplatin was almost always associated with bleomycin and Vinca alkaloids making imputability difficult. Agents targeting the VEGF‐VEGFR axis are associated with hypertension, thromboembolic events and induce microvascular rarefaction 89, but their use is not associated with RP and peripheral vasoconstriction.

Among other cancer chemotherapies that could be responsible for RP, there is limited evidence for gemcitabine 90, 91, 92, vincristine 93, 5‐fluorouracil 94, oxaliplatin 95, tegafur and uracil 96 and cyclophosphamide‐methotrexate‐5‐fluorouracil adjuvant therapy 97.

The pathophysiology of RP induced by cancer chemotherapies is not well understood and is probably multifactorial. Some studies showed an exaggerated response to cold not only in patients with RP but also in patients without finger symptoms before testicular chemotherapy 74, 98. An increased central sympathetic vasoconstrictor reflex and an impaired non‐neurogenic vasomuscular, auto‐regulation was highlighted in patients suffering from RP syndrome after chemotherapy when compared with the control group (patients without RP after chemotherapy) 99. Currently, one of the main mechanisms proposed is through the vascular damage induced by chemotherapy, i.e. endothelial dysfunction that persists after chemotherapy 75. Indeed, some authors 100 showed that microalbuminuria, considered to be a sign of endothelial damage, was significantly higher in patients who received testicular cancer chemotherapy 83. Another possible mechanism is the neurotoxicity of chemotherapies toward arteriolar tone regulation, particularly through hypomagnesia related to cisplatin administration leading to dysregulation of vascular smooth muscle tone 101. RP would appear at the same time as the tubular damage. It is interesting to note that bleomycin is used to induce a sclerodermic phenotype in animals 102, scleroderma being the main aetiology of secondary RP.

Occupational and/or environmental exposure

For some time vinyl chloride exposure has been linked to RP 103. The vascular endothelial toxicity of vinyl chloride has been shown by angiographic studies of arteries in the hand and by capillaroscopy 104, 105. The prevalence of RP in vinyl chloride workers ranges from 6 to 33% 106. In 1980 a prospective exposed/non‐exposed cohort study showed a strong association between vinyl chloride exposure and RP (P<0.006) 107.

Drugs increasing blood viscosity and enhancing vasoconstriction

Interferons (IFN)

RP is a known side effect of treatment with interferon supported by numerous cases reports 41, 108, 109, 110, 111, 112, 113, 114, 115, 116. On direct questioning of patients taking IFN 117, symptoms of RP were reported by more than half. Analysis of 24 case reports of RP associated with interferon 118 highlighted that IFNα is the most common substance implicated (n = 14), followed by IFN γ (n = 5) and IFN β (n = 3). The treatment period was variable and lasted from 2 weeks to 49 months (mean 15.5 months). Clinical findings varied from mild and transient vasospasm (1 h after injection) to digital necrosis in 14 cases. Outcomes were known for 15 patients. Spontaneous recovery occurred for 50% of them after withdrawal of the drug. The remaining patients needed specific medication and six amputations were necessary, underlining the severity of this adverse reaction. A recent meta‐analysis 119 with six eligible studies and 183 patients estimated the prevalence of RP in patients taking interferon to be 13.6% (95% CI 0.026, 0.313).

Currently, the pathophysiology of this reaction is not fully understood. However numerous hypotheses have been proposed: a direct vasospastic effect 113, 120, increasing levels of intracellular fibroblast growth factor in endothelial cells leading to proliferation of these cells and increasing angiogenesis 121 and induction or exacerbation of a dormant collagen disease 109. Although some case reports of RP induced by interferons are described without any immune deficiency 122 it is known that interferon therapy can be related to an autoimmune disease 123, 124. Increasing blood viscosity by induction of serum cryoprecipitation 125, deposition of immune complexes 126 and arterial occlusion by thrombi due to the procoagulant activity of interferon 112, 113 have been proposed. A study of 108 patients with SSc found a higher level of IFN‐γ in patients with associated RP and suggested a pathogenic role of INF‐γ in SSc patients with RP, but this role still remains unclear 127.

Unknown mechanisms

Tyrosine kinase inhibitors

The relationship between tyrosine kinase inhibitors (TKI) and RP is complex. Experimental studies have shown that receptors with a tyrosine kinase activity may play a role in the exaggerated vasoconstriction in response to cold 128. On one hand, a pilot study 129 that included three SSc patients treated with 100 mg day^–1 of imatinib for 6 months showed improvement of their RP 86. Indeed in each patient, RP was attenuated at around 3 months and had completely disappeared at 6 months. On the other hand, exactly the opposite reaction has been described with other TKIs. Emergence of RP during the first week of treatment with nilotinib has been described in two patients 130. One of them experienced improvement after the treatment was switched to imatinib, with recurrence of RP on the reintroduction of nilotinib. Another patient experienced recurrent RP with nilotinib 131. Erlotinib had also been implicated in the case of a 72‐year‐old patient suffering from scleroderma and secondary RP who experienced digital necrosis 20 days after starting daily oral treatment of 150 mg 132. Erlotinib was promptly discontinued and treatment with calcium channel blockers, nitrates and anti‐platelet drugs was initiated. After 3 weeks of therapy, the digital lesion was completely healed. Erlotininib was scored as producing a probable adverse drug reaction (7/10 on the Naranjo scale).

Other

In the literature sporadic case reports of RP potentially induced by drugs can be found, such as the two cases of fluorescein induced RP 133, 134, sulfasalazine 135, 136, propofol 137 and amphotericin B 138 without being able to determine the pathophysiological mechanism. Some paradoxical reactions following the repeated administration of iloprost 139 or yohimbine 140, a selective α₂ adrenergic antagonist, have even been described.

To our knowledge, no pathophysiologic mechanism has been identified yet.

Conclusion

RP is complex, multifactorial and not fully understood yet. This present review summarises the prevalence and level of evidence of the association between drugs and RP (Table 1). Microvascular impairment is a key feature of its pathophysiology. Only symptomatic treatment with vasodilators such as calcium channel blockers or phosphodiesterase‐5 inhibitors has been proposed as a treatment for RP. Logically, vasoconstrictors have long been known to induce or aggravate RP. Increased vascular tone may be related to increased sympathetic activation, but also to endothelial dysfunction or neurotoxicity. Other mechanisms include decreased red blood cell deformability and increased platelet aggregation, both leading to increased blood viscosity (Figure 2). The need for future high quality research including prospective and vascular physiology studies to clarify these mechanisms is obvious. Indeed, this review highlights the lack of available evidence regarding the prevalence of drug‐induced RP, as well as the heterogeneity of its clinical presentation. This probably contributes to the underestimation of drug‐induced RP, as well as the fact that RP is a usually benign condition. However, such an adverse event may rarely lead to serious complications like critical digital ischaemia. Therefore, when these treatments are started in patients with a history of RP, careful monitoring must be made and, if possible, alternative therapies that do not alter peripheral blood flow should be considered.

Table 1.

Most relevant prevalence and level of evidence (defined by the Oxford Centre for Evidence‐based Medicine ‐ Levels of Evidence) of the association between each drug and RP

Mechanism	Drug	Prevalence	Level of evidence
Enhancing vasoconstriction	Clonidine	Unknown	C
	β‐adrenoceptor blockers	7%	A
	Ergot alkaloids	0.1%	C
	Dopaminergic agonists	Unknown	D
	SSRIs	Unknown	D
	Sympathomimetic drugs	Unknown	B
	Ciclosporin	Unknown	B
Endothelial damage	Chemotherapy	20.7–37%	A
	Vinyl chloride	6–33%	A
Drugs increasing blood viscosity and enhancing vasoconstriction	Interferons	13.6%	B
Unknown mechanism	Tyrosine kinase inhibitors	Unknown	D

A: systematic review (with homogeneity) of RCTs or individual RCT (with narrow confidence interval); B: cohort or case control studies ;C: Case series; D: Expert opinion or troublingly inconsistent or inconclusive studies of any level; RCT: randomized controlled trial.

Figure 2.

Schematic representation of some of the key mechanisms contributing to the pathogenesis of iatrogenic Raynaud's phenomenon

Competing Interests

They authors do not have any conflict of interest that might bias the present work.

Khouri, C. , Blaise, S. , Carpentier, P. , Villier, C. , Cracowski, J. ‐L. , and Roustit, M. (2016) Drug‐induced Raynaud's phenomenon: beyond β‐adrenoceptor blockers. Br J Clin Pharmacol, 82: 6–16. doi: 10.1111/bcp.12912.