An Optimal Method for Experimental Provocation of Polymorphic Light Eruption

Background  There is controversy about the best method to induce polymorphic light eruption (PLE) experimentally.

Objectives  To review articles on PLE induction and design a UV radiation protocol that improves success rates with clinically relevant doses of environmentallyrelevant solar-simulated radiation (SSR).

Design and Setting  All articles on the experimental provocation of PLE published since 1980 were reviewed. Photoprovocation of lesions was studied in 25 PLE patients.The 24-hour minimal erythemal dose (MED) of SSR was determined. Thereafter, six 4 × 4-cm adjacent sites on previously affected and previously unaffectedskin were exposed to 0.25, 0.5, 0.75, 1.0, 1.25, 1.5 MED of SSR for 3 to 4 consecutive days. The study periodwas autumn to spring in London, England(51° north latitude).

Main Outcome Measures  Relationship between PLE induction and biological and physical exposure parameters.

Conclusions  The review shows that fractionated erythemally effective UV-A exposures were more successful than single-sunburning UV-B doses. Photoprovocation ofPLE was successful in 68% of patients after 2 to 3 SSR exposures that were not necessarily erythemal. There was no difference in success rate betweenpreviously affected and previously unaffected skin. Our data indicate that PLE is more likely to be induced when the natural causes of the disease aresimulated.

Polymorphic light eruption (PLE) is the most common of the so-called idiopathic photodermatoses. In a recent population survey¹ inthe United Kingdom, PLE was found to affect 15% of healthy people, with a female-male ratio of approximately 2:1. Genetic modeling using twin studiesand families with PLE provides unequivocal evidence for a genetic basis for PLE and predicts that susceptibility to PLE is a polygenic trait with multiplesusceptibility loci.^2,3 Polymorphic light eruption is characterized by a delayed abnormal reaction to the UV radiation(UVR) component of sunlight consisting of transient, nonscarring, pruritic papules and vesicles, typically developing hours or days after sun exposureand resolving over several days without sequelae. The pathogenesis of PLE is unclear, but histologic and immunologic studies suggest that the normalUVR-induced suppression of cell-mediated immunity is impaired and, as a result, these patients develop a T-cell mediated response to a UVR-activated antigen(photoantigen).^4,5

Investigations into the pathogenesis of PLE have been frustrated by the lack of reliable laboratory methods for the induction of clinical lesions.The results of such studies have varied considerably. The early PLE induction studies reported success rates from as low as 10% to 30%^6-8 toas high as 60% to 95%.^9-11 These studies were based on one or more exposures of a single test site to between1 and 10 times the minimal erythemal dose (MED) of nominal UV-B radiation (ie, 290-320 nm), although these sources almost certainly contained otherUVR wavelengths. Initially, induction was achieved with UV-B,^6-13 but several authors also succeeded in reproducing PLE lesions with UV-A irradiation(320-400 nm).^12,14-21

There are somewhat contradictory results about the wavelength dependence of PLE. In general, UV-A appears to be more effectivethan UV-B in eliciting lesions.¹⁴ In a comprehensive study of 142 patients in which induction was attempted with increasing exposuresof buttock skin to UV-A and/or UV-B daily for 4 to 8 days, the most effective waveband was UV-A (56%), followed by UV-A/UV-B combination (27%), then UV-B>(17%).¹² However, Miyamoto¹³ has confirmed earlier reports^22-24 that UV-B can also be successful in a high proportion (57%) of selected patients.In a retrospective study of 30 patients by Mastalier et al,²¹ the action spectrum fell within the UV-A range in 59% of patients, the UV-B range in 23%, and both ranges in 18%. The diversity in the wavelength-dependence studies remains unexplained. It would be puzzling if a single chromophoreand a uniform mechanism could be activated by different parts of the UVR spectrum and result in such a morphologic diversity of lesions. This would indicatedifferent pathogenic mechanisms or that there is a range of UVR-induced antigens.

According to Norris et al,²⁵ studies of laboratory-induced lesions have generally used UVR doses considerably inexcess of the MED.^26-28 This makes histologic interpretation difficult because of coexistent sunburn, whichresults in mononuclear cell infiltration. Even recent studies have used high doses to provoke a PLE reaction (3.3 MED of UV-B¹⁹;6 MED of UV-B²⁰; and 2-4 MED of UV-B or UV-A²⁹).

Different locations have been used to provoke PLE with varying degrees of success. Barnadas et al,¹⁸ McFadden et al,³⁰ Mastalier et al,²¹ andNorris et al³¹ used previously affected skin with success rates of 30%, 100%, 57%, and 100% respectively. Boonstra et al²⁹ provoked PLE in previously affected skin to UV-B/UV-Acombination exposure in 88% of men and 52% of women. Within this group, PLE was also induced by UV-B alone in 9% of the men and 24% of the women. Reactionsto UV-A alone occurred in 3% of men and 24% of women and to visible light in 43% of men and 11% of women. Verheyen et al¹⁹ andLambert et al²⁰ used whole body and upper arm exposure and had UV-A success rates on previously affected sites of 100% and87%, respectively, and 0% and 6.7% with UV-B. Holzle et al¹⁴ tried to provoke PLE on previously exposed and previously unexposed skin. The successrates ranged from 0% to 90% on previously exposed skin, but it was 0% on previously unexposed skin.

A synopsis of findings from our literature review is provided in Table 1. The aim of our investigation was to evaluate provocation test methods introduced since 1980 and to developa reliable test using physiologically relevant doses of SSR to study the pathogenesis of PLE.

Twenty-five patients diagnosed with PLE at the Photobiology Clinic of St John's Institute of Dermatology, London, England, were randomly recruitedinto the study (21 women, 4 men; age range, 18-55 years; mean age, 38 years; skin types I-V). Diagnosis was made on the basis of clinical history and examination.All patients had PLE for at least 6 years, and the condition was defined as a fully resolving macular, papular, or papulovesicular photoeruption occurring30 minutes to 48 hours after sun exposure and lasting for hours to at least a day.

Patients were screened by blood analysis to exclude those with porphyria and by estimation of antinuclear and extractable nuclear antigen autoantibodiesto exclude those with lupus. Skin type was assessed by interview. Exclusion criteria for all volunteers included ingestion of any medication during or2 weeks prior to the study (oral contraceptives excepted), phototherapy during or in the 6 months prior to the study, previous exposure of buttock skin tosunlamps or sunlight, and/or a recent sunburn. Pregnant or lactating women were also excluded. The local ethics committee approved the study, and eachvolunteer was fully informed of the procedures and gave written informed consent to participate.

Solar-simulated radiation (SSR) was generated by a 1-kW xenon arc solar simulator (Oriel, Stratford, Conn) fitted with a WG320/1-mm-thick glass filter,giving an even field of irradiance (approximately 290-400 nm) of about 15 mW/cm² on the skin surface at 11 cm from the source. Irradiancewas routinely determined with a wideband thermopile radiometer (Medical Physics, Dryburn Hospital, Durham, England) calibrated against a DM150 double monochromatorBentham spectroradiometer (Bentham Instruments, Reading, England). See filter 2 of the first figure in Harrison and Young³² foremission spectrum details. Eighty-eight percent of the erythemally effective energy of the source was in the UV-B range, and the remaining 12% was UV-A.

In each patient, the just visibly perceptible 24-hour MED was determined on previously unexposed buttock skin using eight 25% doseincrements. Photoprovocation was attempted on 2 body sites: (1) previously affected skin (either the arm or upper back) and (2) previously unaffectedskin (buttock). Six adjacent sites (4 × 4 cm each), were exposed to0.25, 0.5, 0.75, 1.0, 1.25, 1.5 buttock MED in an attempt to induce PLE lesions.The exposures were repeated daily until a positive PLE response was obtained. If no lesions appeared after 3 or 4 days of exposure, the test was considerednegative. (Twenty-three of 25 patients had a maximum of 3 exposures; only 2 of 25 patients had a maximum of 4 exposures.) Photoprovocation in 18patients was performed from October 2001 to May 2002 and in a further 7 patients from November 2002 to March 2003.

Using a reflectance meter (Diastron, Andover, England), we took 3 quantitative measurements of erythema 24 hours after the second SSR exposure and calculatedthe mean per site. For each site, the increase in erythema index was calculated by subtracting the mean background reading of adjacent nonirradiated skin.

Microsoft (Redmond, Wash) Excel 2000 was used for statistical analysis. Linear regression was performed for increase in erythema index vs SSR dose(MED fraction) for each volunteer on previously exposed and previously unexposedtest sites to generate erythema dose-response slopes (83% of the regressionshad R² values ≥0.7), the slope being indicative of the overall erythema response. The distribution of the slopeswas shown to be normal (Kolmogorov-Smirnov test), and so unpaired t tests were used to compare the slopes between patients who showed a PLE response and those who did not on both treatment sites. Significancewas assumed at P<.05.

The morphologic features of PLE lesions induced by photoprovocation were consistent with the response reported by each patient after previoussun exposure. Thus, with SSR we were able to produce lesions comparable to those seen in the genuine disease. The PLE lesions developed 1 to 24 hoursafter the last irradiation and consisted of macular, small papular (Figure 1A), or papulovesicular (Figure 1B) lesions scattered over the irradiated site. The papular or papulovesicular lesions were sometimes sparsely scatteredover the irradiated area, although almost confluent lesions developed as well.

In all patients, the recurrence of the rash was associated with pruritus. Shortly thereafter, patchy or confluent erythema developed in which the characteristicskin lesions emerged. Lesions persisted for at least 12 hours and up to 2weeks after the last exposure. Figure 1Ashows a papular reactions of increasing severity after 3 consecutive exposures to 0.75, 1.0, 1.25, and 1.5 MED. Figure 1B shows a papulovesicular PLE reaction after 3 consecutive exposureson 1.25 and 1.5 MED sites. Experimentally induced skin lesions usually resolved over the next 1 to 3 days, but it generally took 1 to 3 weeks before theycompletely disappeared.

Erythema was shown to be SSR dose dependent, and the slopes were significantly steeper on the buttock site than on the previously exposed site (Figure 2). There was no difference in theerythema response on either test site between patients in whom PLE was successfully induced and those in whom it was not (Figure2).

Seventeen (68%) of 25 patients developed PLE on at least 1 test site. Figure 3 showsthe number of sites with a positive response at a given dose level, regardless of the number of exposures. Six patients developed PLE only on previously affected sites (upper back or arm); 4 patients developed PLE only on previouslyunaffected sites (buttock); and 7 patients showed responses on both sites.Polymorphic light eruption was readily provoked on both previously exposed and previously unexposed skin. Eleven (65%) of 17 patients responded on buttockskin and 13 (77%) of 17 responded on arm and/or back.

Successful provocation of PLE was dependent on the number of exposures.In both skin sites, success increased with increasing number of daily exposures. Figure 4 shows the cumulative success ratefor both sites, regardless of the dose used. The data show that 3 exposures are necessary to obtain a success rate of more than 50%. However, Figure 3 also shows that it is possible toprovoke PLE with a single SSR exposure on buttock skin (1 patient with macular lesions) and on arm and/or back skin (2 patients with papular lesions).

A summary of articles published since 1980 on the experimental provocationof PLE is provided in Table 1.This shows a very diverse range of UVR spectra, doses, and induction protocols, and sometimes it is difficult to determine the exact radiation protocol. Ingeneral, the doses given, whether single or repeated, have been greater than 1 MED. Only our group²⁵ has previously usedSSR in a suberythemal protocol. Furthermore, in some cases the doses would have been phototoxic, causing a severe erythemal reaction.

In general, studies with UV-A have been more successful than those with UV-B. Solar UVR is mostly UV-A (approximately 95%), yet the approximate 5%UV-B causes more than 80% of the sunburn. Patients often report PLE without sunburn, which has led clinicians to believe that the disease may be triggeredwith suberythemal UV-A exposure. Yet, success with UV-A dose-response studies has been with doses that were certainly erythemal (eg, 60-100 J/cm² in Holzle et al¹⁴ and >50 J/cm² in 20 of 22 patients in McFadden and Larsen¹⁵).In this context, such studies do not reproduce the natural situation.

We have successfully reproduced PLE in a small group of PLE patients using environmentally and physiologically relevant UVR exposures. This wasdone with comparable rates of success on previously exposed and unexposed skin. The erythemal response after 2 exposures on the buttock was much greaterthan that on the arm and/or back, as shown in Figure 2, almost certainly because buttock MED was lower than the MED of sites that had a history of exposure. In other words, our protocolwas less inflammatory on the arm and back than on the buttock.

Polymorphic light eruption can be induced by suberythemal exposures, as shown in Figure 3. Figure 2 shows PLE in association with very shallow erythema dose-responsecurves, especially on the arm and back. These observations are consistent with patient reports that PLE may evolve without concomitant sunburn. Overall, our data show that erythema and PLE are independent clinicaloutcomes, suggesting that they have different chromophores. DNA is thought to be the major chromophore for erythema,³³ butas yet we have no candidate chromophores for PLE.

The most frequent outcome was pruritus and erythema in combination with papular or papulovesicular lesions, which accounted for 88% of the positivereactions. The remaining 12% involved pruritis, erythema, and macules that persisted for at least 12 hours after the last exposure. The lesions weresometimes sparsely scattered over the irradiated area, which means that test areas should not be too small. Thus, we believe that our 4 × 4-cm testarea for each dose is appropriate. Larger areas are not necessary: Neumann et al¹⁶ have shown that whole body UV-A exposurewas no more effective than the exposure of smaller previously involved skin sites.

Our success rate was 68% (17/25), which is good compared with most of the reports in Table 1. Therewas little difference between the success rates on previously affected sites(13/17; 76%) and previously unaffected sites (12/17; 71%). Most studies haveattempted to induce PLE on previously sun-exposed sites.^{12,14,15,18,31} Only 2 other studies^14,18 attempted,without clear success, to induce PLE on previously unaffected (ie, non–sun-exposed) sites. Our results clearly show that successful provocation of PLE is notdependent on skin site.

Polymorphic light eruption usually requires several consecutive exposures to sunlight. We thus tried to simulate the natural conditions with repeateddaily exposures of 0.25, 0.5, 0.75, 1.0, 1.25, and 1.5 just perceptible MEDs. Unfortunately, the total number of exposures was limited because most patientswere not willing to cooperate for more than 4 consecutive exposures. The influence of number of exposures is shown in Figure4. This shows that at least 3 exposures are necessary to obtaina success rate of at least 50% on previously exposed and previously unexposedskin.

The SSR dose (cumulative physical dose) required to induce PLE ranged from 1.7 to 25 J/cm² (mean MED, 4.9 J/cm² ; range, 2.6-10.1J/cm²), and there was no evident correlation between outcome and cumulative physical dose. In a recent study,⁵ membersof our group used single exposure doses of 0.6, 1, or 2 MED to induce immunosuppression in PLE patients and in age- and skin type–matched controls. Exposureto 1 MED suppressed contact hypersensitivity response by 78% in controls but induced significantly less suppression (44%) in PLE patients (P<.01). Suppression was also less in PLE patients than in controlsafter 0.6 MED (31% vs 43%), but this did not reach significance (P = .50). In contrast, suppression was almost complete (93%) in both PLE patients and controls after exposure to 2 MED, suggesting that PLE patientshave a resistance to immunosuppression after low to moderate doses of UVR, but resistance to immunosuppression can be overcome if higher UVR doses aregiven.

It is possible that more exposures would have increased our success rate. On the other hand, the clinical symptoms of PLE are likely to representa balance between the induction of the putative photoantigen and the UVR-induced suppression of the immunologic response to the antigen. Such a balance islikely to be highly dependent on individual immunologic and UVR exposure parameters. Polymorphic light eruption was provoked after a single exposure in 3 (18%)of 17 patients, and in 1 of these cases the PLE was evident after the MED series. This contrasts with the findings of other authors,^{15,25,27,28,30,31} whoreported good results from a single exposure with a wide range of sources and doses.

In conclusion, we have shown that PLE can be readily induced when thenatural causes of the disease are simulated. The lack of consistent resultsfrom studies of the PLE action spectrum further supports the use of SSR. We have also shown, for the first time, that the condition can be just as readilyprovoked on previously unexposed sites. This indicates that the putative photoantigencan be induced de novo. Finally, we believe that a standard protocol for theinduction and assessment of PLE would benefit research and clinical practice, and we advocate that researchers in this field work toward this end.

Corresponding author and reprints: Antony R. Young, PhD, St John'sInstitute of Dermatology, Photobiology Unit, King's College London, SecondFloor South, Wing Block 7, St Thomas' Hospital, London SE1 7EH, England (e-mail: antony.r.young@kcl.ac.uk).

Accepted for publication June 25, 2003.

This study was supported by a personal training award from the Guy's & St Thomas' Charitable Foundation, London, England (Dr van de Pas), anda project grant from the Claneil Foundation Inc, Plymouth Meeting, Pa.