Systemic drug photosensitivity—Culprits, impact and investigation in 122 patients

Systemic drugs are a potentially reversible cause of photosensitivity. We explore prevalence, impact, phototest findings and culprit drugs.


| INTRODUC TI ON
Drug-induced photosensitivity is an important, potentially reversible cause of photosensitivity, and is a potential adverse effect of many medications. While systematic reviews have explored the range of medications causing photosensitivity, 1,2 the prevalence of these drug reactions is largely unknown. A retrospective analysis of reports suggested drug-induced photosensitivity may account for up to 8% of cutaneous adverse effects from drugs. 3 However, estimates of prevalence are likely to be an underestimate as many cases may be undiagnosed or unreported. Studies are largely based on clinical observation, including case series and case reports, with a minority incorporating phototesting. 2 Objective evidence is therefore usually lacking. Moreover, culprit drugs and the expression of drug-induced photosensitivity alter as new drugs emerge. Thus, more awareness is required regarding current culprits, diagnosis and impact on patients.
Mechanisms of drug-induced photosensitivity include phototoxicity, photoallergic reactions and drug-induced lupus erythematosus (LE), 4 with most cases of systemic drug-induced photosensitivity thought to be exerted through phototoxicity. Pathogenesis of phototoxic reactions is secondary to activation of the photosensitizing drug or its metabolite(s) by ultraviolet radiation (UVR) which can then lead to photosensitivity either through direct cellular or oxidative and free radical damage. 4,5 Phototoxicity may theoretically occur in any individual exposed to enough of the drug or relevant wavelength of UVR, although the threshold differs across individuals. 4 Photoallergic reactions to systemic drugs are less common, more often occurring due to topical agents. They are mediated by cell-mediated type IV hypersensitivity responses and require sensitization to the offending drug. 5 With topical though not systemic photoallergy, photopatch testing is usually very useful in determining the causative agents.
Patients with drug-induced photosensitivity can be investigated at a photodiagnostic unit, where their features of photosensitivity are differentiated from those of other photodermatoses. A detailed clinical appraisal is required to explore for culprit drugs, including timing of onset of symptoms and of medication. Clinical features may include burning, itching and a rash affecting sun-exposed sites; this is most often either with a pronounced sunburn-like reaction or with an eczematous rash, while other features including photo-onycholysis, LE and lichenoid reactions occur more rarely. Monochromator phototesting to examine erythemal thresholds to narrowband UVB, UVA and visible radiation helps objectively identify presence of photosensitivity and its action spectrum. 4 Resolution of abnormal responses on retesting or clinical improvement after a several month period of drug cessation can assist confirmation of drug-induced photosensitivity.
Our objectives were to explore the prevalence, impact, current and emerging culprits of drug-induced photosensitivity, in patients with features of photosensitivity investigated at a photodiagnostic unit.

| Patients and methodology
This was a case series of patients diagnosed with drug-induced photosensitivity over a 16-year period (Jan 2000-Jan 2016) in a specialist clinical photoinvestigation centre (Photobiology Unit, Dermatology Centre, Salford Royal NHS Foundation Trust, Manchester, UK).
Patients are referred, mainly by general dermatologists, from a wide geographic area of Northern and Central England and Wales, UK.
Data were obtained with standardized pro forma and phototest methods. Detailed clinical history was taken from patients with timelines of medications taken and symptoms of photosensitivity and they were examined for signs of a photodistributed rash and its morphology. Detailed monochromator phototesting was performed to narrow bandwidths of UVR and visible radiation from 300 to 600 nm. Additionally, patients underwent broadband phototesting with low-dose UVR on up to 3 consecutive days in accordance with the unit's routine photo-provocation schedule. Photopatch with control patch testing was performed to sunscreen filters and non-steroidal-anti-inflammatory drugs (NSAIDs), along with laboratory tests including connective tissue disease (CTD) screen, urine and blood porphyrin testing and serum 25-hydroxyvitamin-D (25OHD) level.
Dermatology Life Quality Index (DLQI) was evaluated, with scores for the past week and the past year. 6

| History and examination pro forma
Detailed history and examination were taken using a standardized pro forma to include: age and season at onset; number of episodes per year; whether rash is continuous or present/absent/less severe in winter; any increased tolerance as summer progresses; precipitants of rash; use of sunscreens and whether use helps prevent rash; whether rash can be provoked through window glass; duration of sun exposure needed to provoke rash; duration between exposure to sun and appearance of rash; duration of lesions; symptoms of condition; drug history at the onset of photosensitivity with start and stop dates; present systemic drugs and topical agents with start and stop dates; any significant consumption of food/drink containing psoralens or quinine; previous treatments for the rash including dates; past medical history; family history; phototype; occupation/ outdoor activities; effect on quality of life (including restriction in outdoor activity). Examination of rash and available photographs was performed to evaluate morphology and distribution.

| Monochromator phototesting
Phototesting against narrow bandwidths of UVB, UVA and visible radiation was performed on the central/upper back to determine the minimal erythemal dose (MED) and define the action spectrum. Patients were exposed to 300, 320, 330, 350, 370, 400, 500 and 600 nm (half-maximum bandwidth 5 nm at 300 nm, 10 nm at 320 and 330 nm; 20 nm at all other wavelengths) using a xenon arc lamp (1 KW short arc, Newport Spectra-Physics Ltd, Didcot, UK) coupled to a 1/4m grating monochromator (Newport Spectra-Physics Ltd) ( Table 1). Irradiance was measured using a calibrated thermopile (Medical Physics, Dryburn Hospital, Durham, UK) and digital voltmeter (Medical Physics, Royal Liverpool University Hospital, Liverpool, UK). Patients were exposed on their back to a geometric progression of doses (common ratio = √2) at each wavelength. The lowest dose at which a visible erythemal reaction was induced, that is the MED, was noted at 24 hours. 7

| Broadband UVR provocation testing
Testing was performed on 5 × 5 cm areas of skin of the ventral forearm. All patients were given 15 J/cm 2 broadband UVA (320-400 nm) using a custom-built arm exposure unit incorporating Cleo Performance™ bulbs (Phillips Healthcare UK Ltd. Guildford UK).

| Photopatch testing
This was carried out on the lower back to examine for presence or co-existence of photocontact allergy. From 2009, this involved 24 agents: 19 UV filters, 5 NSAIDs (Chemotechnique Diagnostics, Vellinge, Sweden) 8 and sunscreen products including patients' own, while prior to 2009, the photopatch series comprised 10 agents: 9 organic UV filters and sunscreen products. Duplicate patches were applied (day 1) to skin of the mid-back for 24 hours following which one set was irradiated (day 2) with broadband UVA (5J/cm 2 ; reduced dose where low erythemal thresholds) (320-400 nm; UVAL 801, Herbert Waldmann GmbH & Co. KG, Villingen-Schwenningen, Germany). Visual readings were taken at 24 and 48 hours post-UVR (days 3, 4) to examine for a response, using the International Contact Dermatitis Research Group (ICDRG) grading. 7,9

| Laboratory tests
These included CTD screen, urine and blood porphyrin testing, and serum 25OHD level.

| Quality of life evaluation
The DLQI is a questionnaire used to assess the quality of life (QoL) which has been validated in dermatology conditions, although not specifically for photosensitivity conditions. 6 The DLQI was completed at the time of photoinvestigation by patients from January 2011 onwards, to assess the impact of drug-induced photosensitivity on QoL both in the past week and past year to take into account seasonal variation.

| Clinical features
The clinical features presented by patients are shown in relation to their suspected culprit drug ( Table 2). The median duration of symp-

| Monochromator phototesting
Patient phototesting results are shown in relation to their culprit drug in Table 3. Overall, 45.1% (n = 55) of patients had reduced (abnormal) MED on monochromator phototesting: 83.6% (n = 46) showed reduced MED to monochromator testing with UVA (320-400nm), 14.5% (n = 8) to UVA plus UVB (300-400 nm) and 1.8% (n = 1) to UVA plus visible light (320-500nm). In patients with UVA sensitivity and accompanying UVB and/or visible light sensitivity, MED was reduced to a greater degree in the UVA than UVB or visible light regions.
Nineteen (15.6%) of the 122 patients were phototested after the photoactive drug had been discontinued. The frequency of reduced MED seen at each wavelength on monochromator phototesting in patients diagnosed with drug-induced photosensitivity is shown in Figure 1. broadband UVA only n = 4). Thus, SSR testing was particularly successful, while there was additional benefit from performing both.

| Photopatch testing
Positive photopatch testing to one or more agents with negative control patch testing was seen in 6 of the 122 patients. Agents were ketoprofen 1% (n = 2), butylmethoxydibenzoylmethane/avobenzone (n = 2) and benzophenone-3/oxybenzone (n = 3). In one patient who had a positive photopatch test to benzophenone-3, a relevant previous reaction to sunscreen was reported by the patient and sunscreen photocontact allergy concurrent with the systemic drug photosensitivity was clinically suspected. Additionally, 11 patients showed positive patch testing alone; agents were homosalate, methylene bis-benzotriazolyl tetramethylbutylphenol (Tinosorb M), benzophenone-3, octylmethoxy-cinnamate, benzophenone-4 and patients' own sunscreen products. These findings were unrelated to the patients' systemic drug photosensitivity and patients were advised to avoid products containing these compounds.

| Laboratory testing
Porphyrin plasma and urine test were positive in 2 of 117 patients.
These results were weakly positive urine porphyrin: creatinine ratio in one patient and weakly positive plasma porphyrin peak at 622nm in another patient, which were not considered relevant.  Symptoms resolved on cessation of mesalazine, with normalization of MED. A further patient had positive anti-dsDNA (71 iu/mL), but this was pre-existing in relation to chilblain lupus.
Extractable nuclear antigen (anti-Ro/SSA) was positive in one patient with severe UVA sensitivity, who was felt to have drug-induced LE. Clinical features were severe erythema and oedema affecting sun-exposed sites and were replicated on provocation testing. The patient was taking methyldopa, which is associated with drug-induced LE. However, the onset of symptoms was closely associated with commencement of carbamazepine, which has also been linked to drug-induced LE in case reports, and this was concluded to be the more likely culprit. Anti-La/SSB antibodies were slightly elevated (1.

| Culprit drugs
Suspect drugs are shown (Tables 2 and 3 Azathioprine-induced photosensitivity was also seen in an 11-yearold girl with skin type II. Monochromator phototesting revealed reduced MED to UVA (320-370 nm) wavelengths. Two further cases of azathioprine-induced photosensitivity were seen: a 25-year-old woman, skin type II, exhibiting UVA sensitivity at 370 nm, and a 53-year-old man, skin type II, with abnormal response to broadband UVA and SSR.

TA B L E 3 (Continued)
drug if able with guidance of their relevant specialist or general practitioner. In addition, patients were advised on detailed photoprotection measures, including suitable clothing and hats, the use of high UVA protection and sun protection factor (SPF) sunscreens and UVR blocking window films for car, office and home. Vitamin D supplementation of 800 IU daily was recommended in the case of vitamin D deficiency or insufficiency.
Sixty patients re-attended the photobiology unit, the majority reporting complete resolution of symptoms after cessation of the culprit drug, while 14 reported persistent symptoms. Of the latter, 3 patients were unable to stop the suspect drugs (omeprazole, ciprofloxacin/ramipril, lansoprazole) for medical reasons. A further patient (on bendroflumethiazide), was diagnosed to have photoaggravated eczema. Of the remaining ten patients, photosensitivity resolution was clinically incomplete (suspect drugs: simvastatin n = 2, lansoprazole n = 1, omeprazole n = 1, quinine n = 2, indapamide n = 1, candesartan n = 1, fluoxetine n = 1, sulphasalazine n = 1); those taking additional photoactive drugs were then advised to visit their regular physician to discuss potential cessation/ substitution of these.
Additionally, phototesting remained abnormal in the patient taking ciprofloxacin/ramipril who was unable to stop these drugs.

| Dermatology life quality index
The DLQI questionnaire, introduced routinely in 2011, was completed by 58 patients at the time of photoinvestigation (Table 4). A higher impact was seen for the past year than the past week, with many patients visiting the photoinvestigation unit in winter time.
Median DLQI score was 6 (range 0-29) for the past week and 11 (range 2-27) for the past year.  24 Of concern, azathioprine is also associated with photocarcinogenesis; this is thought to be exerted through metabolites such as 6-thioguanine causing promutagenic oxidative DNA damage following interaction with UVA. 24,25 Our series also revealed one case of etanercept-induced photosensitivity to UVA, and a further case of possible potentiation of UVA and UVB photosensitivity to azathioprine by infliximab. TNF-α inhibitors have been associated with a variety of cutaneous reactions, including psoriasis, sebopsoriasis, pustulosis, eczematous, bullous, granulomatous, lichenoid and vasculitic rashes, 26 and have also been widely associated with drug-induced LE, 27,28 which may contribute to photosensitivity. However, no similar reports of severe photosensitivity as seen in our cases have, as far as we are aware, been reported.

| D ISCUSS I ON
We found one case of photosensitivity related to denosumab, an anti-RANKL biologic administered subcutaneously to treat osteoporosis. Although there is scarce mention of denosumab-induced photosensitivity in the literature, FDA reports have linked denosumab to photosensitivity; this was seen in 6 treated patients versus 1 on placebo in a phase III randomized clinical trial involving 7800 postmenopausal women with osteoporosis. 29 In recent years, cases of photosensitivity have been reported with newer drugs including cancer therapies such as BRAF kinase inhibitors, for example vemurafenib and dabrafenib, epidermal growth factor receptor (EGFR) inhibitors and other anticancer therapies. 4 No cases were seen in our photoinvestigation centre in relation to these agents, perhaps due to awareness and management of the cutaneous adverse effects by the oncology community.
What makes certain patients more likely to develop drug-induced photosensitivity is unclear. It is probable that genetic factors, such as polymorphisms in genes encoding drug-metabolizing enzymes, antioxidant properties and sensitivity to UVR play a role. 4 We found the gender balance of patients to be equal (female 52.5%, n = 64). Interestingly, the yield of abnormal responses was increased by the addition of broadband phototesting; this might be anticipated in view of the broad action spectrum demonstrated by several phototoxic drugs. 4 The majority of lowered MED were to UVA alone, and while a number occurred to UVB in addition to UVA, none were seen to UVB alone (Table 3). Only one patient, taking bendroflumethiazide, showed lowered MED to UVA plus visible light. This action spectrum was anticipated, that is reflecting the frequency of UVA > UVB > visible light activation. 4 Limitations of this study include the challenges in identifying culprit drugs in patients who take several medications and where a concurrent photodermatosis, such as photoaggravated eczema, may confound presentation. There is no definitive diagnostic test for drug photosensitivity, and consequently, this diagnosis is inevitably largely presumed rather than proven. It would be optimal for photodiagnostic units to follow up all patients suspected of drug photosensitivity, including repeat phototesting, although this is often not practically possible. All our patients are offered follow-up appointments, but as many travel long-distance to the specialist unit, they often arrange to return only if their symptoms persist, deferring their appointment if they resolve.
In conclusion, we find drug-induced photosensitivity occurred in 5.4% patients referred to a specialist photoinvestigation unit and suspect a much larger population of patients fails to be suspected and referred. Classical photosensitizing drugs such as quinine and thiazides remain common culprits; physicians should be aware of these as well as emerging culprits including the biologics, PPI and statins. Drug-induced photosensitivity causes very high impact on QoL and is vital to suspect in view of its potentially curable nature.
Early suspicion, identification, consideration of phototesting/repeat phototesting alongside cessation of culprit drugs where feasible, and implementation of photoprotection, are key measures.

ACK N OWLED G EM ENTS
The study was funded by the Kuwaiti Government and the National Institute of Health Research Manchester Biomedical Research Centre.

CO N FLI C T O F I NTE R E S T
None declared.