Azathioprine is associated with enhanced skin photosensitivity to ultraviolet A (UVA) and leads to incorporation of 6-thioguanine (6-TG) into DNA of dividing cells. Unlike canonical DNA, 6-TG DNA is damaged by UVA, which comprises more than 90% of the ultraviolet reaching earth. Skin photosensitivity to UVA and UVB was measured in 48 kidney transplant patients immunosuppressed either by azathioprine (n = 32) or mycophenolate (n = 16). In 23 patients, azathioprine was subsequently replaced by mycophenolate and skin photosensitivity, DNA 6-TG content in peripheral blood mononuclear cells, and susceptibility to UVA-induced DNA damage were monitored for up to 2 years. The mean minimal erythema dose to UVA on azathioprine was twofold lower than on mycophenolate. Three months after replacing azathioprine by mycophenolate mofetil, the minimal erythema dose to UVA had increased from 15 to 25 J/cm2 (p < 0.001) accompanied by reduced DNA 6-TG content. P53 protein expression in irradiated skin indicated reduced susceptibility to UVA-induced DNA damage. 6-TG DNA in peripheral blood mononuclear cells remained measurable for over 2 years. Replacing azathioprine selectively reduced the skin photosensitivity to UVA, attenuated UVA-induced skin DNA damage, and is likely based on incorporated 6-TG in DNA.
More than half of all organ transplant recipients develop skin cancer—in particular, squamous and basal cell carcinoma (1). Immunosuppressive drugs (2) and cumulative exposure to sunlight (3) contribute to this increased risk for skin cancer. Ultraviolet radiation from sunlight can interact with certain drugs to increase skin photosensitivity (4). The adverse skin effects that ensue from the combination of photosensitizing agents and sunlight have similarities with sunburn—an acknowledged risk factor for skin cancer (5).
The immunosuppressant azathioprine (AZA) is widely used in organ transplant recipients. AZA is activated to thioguanine nucleotides that are precursors for the incorporation of the base analog 6-thioguanine (6-TG) into DNA (6). One important property of DNA 6-TG is its ability to absorb the ultraviolet A (UVA) radiation, which comprises more than 90% of the ultraviolet incident at the earth's surface. Canonical DNA bases absorb UVA very poorly, and these wavelengths do not damage DNA directly and cause only minor damage by indirect mechanisms. The presence of 6-TG converts DNA into a UVA chromophore and the absorbed energy is a source of highly damaging reactive oxygen species (7). Both DNA 6-TG itself and the canonical DNA bases are susceptible to damage by reactive oxygen species. Cultured human cells are extremely sensitive to killing and mutation by photochemical DNA damage caused by the interaction between DNA 6-TG and UVA. Furthermore, AZA increases the risk of ultraviolet radiation-induced skin cancer in hairless albino mice (8).
There are indications that these damaging photochemical reactions are of clinical significance. The skin of patients taking AZA is selectively hypersensitive to UVA (9), although the underlying molecular events and their potential reversibility are not known. We, therefore, studied the relationship between the enhanced skin photosensitivity and DNA damage in long-term kidney transplant recipients.
Adult kidney transplant recipients were eligible for enrolment in the study if they had received the allograft at least 2 years previously, had been on continuous treatment with either AZA or mycophenolate since transplantation, and had a glomerular filtration rate (GFR) of more than 20 mL/min estimated by the Modification of Diet in Renal Disease (MDRD) formula (10).
Long-term kidney allograft recipients were enrolled at the University Hospital Zürich between April 2006 and March 2009. The study was carried out in accordance with the ethical principles of the Declaration of Helsinki, the Good Clinical Practice guidelines of the International Conference on Harmonization, local regulatory requirements and local medical ethics committee approval. All patients provided written informed consent.
The baseline photosensitivity to UVA and ultraviolet B (UVB) was studied in transplant recipients receiving AZA or mycophenolate. In a subsequent exploratory prospective, nonrandomized, open-label trial, patients on AZA were then switched to mycophenolate mofetil (MMF). Photosensitivity was measured at the day of the switch to MMF and 3 months later. All patients on AZA were switched to a dose equivalent of MMF as follows: those on 25 mg/day AZA were switched to 0.5 g/day MMF, those on 50 mg/day AZA to 1 g/day MMF, those on 75 mg/day AZA to 1.5 g/day MMF and those on 125 mg/day AZA to 2 g/day MMF, respectively.
On the day of the switch, at 2 weeks, and at 3 months, serum blood creatinine was measured and the total urinary protein to creatinine ratio (gram per millimol) was determined. The GFR was estimated according to the abbreviated MDRD formula, using the following parameters: creatinine, gender, ethnicity and age (10). Steady-state blood levels of ciclosporin and tacrolimus were measured by immunoassay. Vital signs and adverse events were recorded at each study visit.
The skin in the sacral region was exposed to incremental doses of UVA and UVB. The UVA-source (Supuvasun Mutzhas 3000, Munich, Germany) had a spectral output of 350–450 nm (peak 370–385 nm). This is within the UVA band and contains no UVB. Twelve evenly spaced doses between 2.5 and 49.6 J/cm2 were delivered to a defined surface of approximately 2.5 cm2. UVB photosensitivity was determined using a Waldmann UV-800 source (Waldmann, Villingen-Schwenningen, Germany) with a spectral output mainly within the UVB band (285–350 nm, peak 310–315 nm). The radiation doses for UVB ranged from 8 to 99 mJ/cm2. The doses were delivered in six equally spaced increments. Photosensitivity was defined as the minimal erythema dose (MED)—the lowest dose resulting in just perceptible erythema 24 h after exposure. Instant erythema and pigmentation were also recorded 30 min after exposure. Most patients were examined in the second quarter of the year: 72% and 81% in the AZA and mycophenolate group, respectively. Patients were classified into skin type groups I–V according to the Fitzpatrick classification (11). Skin biopsies were taken from the area of MED on the first and second determination of MED and archived in RNA later.
Determination of DNA 6-TG content in peripheral blood mononuclear cells
DNA was extracted from peripheral blood mononuclear cells using the Wizard Genomic DNA purification kit (Promega, Southampton, UK). DNA 6-TG was quantitatively converted to guanine sulfonate by treatment with magnesium monoperoxyphthalate (1 mM, 30 min in the dark at room temperature). Treated DNA was ethanol precipitated and approximately 50 μg was digested to nucleosides with 10 units of nuclease P1 (1 h, 50°C), followed by 2 units of shrimp alkaline phosphatase (1 h, 37°C). Deoxynucleosides were separated by high-performance liquid chromatography (HPLC) as described (7). Deoxyguanosine was quantified by its A260 and guanine sulfonate deoxynucleoside by its fluorescence. The detection limit for guanine sulfonate deoxynucleoside is 0.1–0.5 pmol.
Expression of the p53 tumor suppressor in skin biopsies
On the day of the switch from AZA to MMF, 4-mm punch biopsies were taken from a site that had received one UVA MED. This was repeated after 3 months on MMF. Paired biopsies on AZA and after 3 months on MMF were obtained from eight patients for immunohistochemistry. Sections (3–5 μm) were prepared and p53 immunoreactive keratinocyte nuclei were identified by immunohistochemical staining. The primary antibody was mouse monoclonal anti-p53 (clone DO7, Cell Marque, Rocklin, CA, USA). Secondary staining was for alkaline phosphatase (12). On each section, three randomly chosen high-power fields were counted for immunoreactive nuclei.
Results are expressed as mean ± standard deviation (SD), median and interquartile range (IQR) or number of patients (percent). For comparisons between groups, means of continuous data were compared by Student's t-test or by the Mann–Whitney U-test, and categorical data by Fisher's exact. Paired data were compared by paired Student's t-test. All p values were two sided for the comparison between the groups or between baseline and follow-up values, and those less than 0.05 were considered statistically significant. Data analysis was performed using SAS 9.2 (SAS Institute Inc., Cary, NC, USA).
A total of 48 kidney transplant recipients were enrolled into the study at the University Hospital Zurich between April 2006 and September 2009 (Figure 1). Twenty-six of 58 patients on AZA and 28 of 44 patients on mycophenolate assessed for eligibility declined study participation. Thirty-two patients were taking AZA (AZA group) and 16 patients mycophenolate (MMF group: 12 patients with MMF and 4 patients with enteric-coated mycophenolate sodium). Most patients were in their fifth decade, had a GFR between 40 and 60 mL/min/1.73 m2, and had a triple immunosuppressive treatment that included ciclosporin and prednisone (Table 1). The majority of patients were of skin type-II and type-III whereas there were proportionately fewer patients of skin type-II in the MMF group (12.5% vs. 31%). History of skin cancer as a reflection of cumulative photodamage of the skin was well matched and indifferent statistically. As expected, however, AZA patients had received their transplant earlier than mycophenolate-treated patients: 15 years and 8 years since transplantation, respectively.
Table 1. Characteristics of kidney transplant recipients according to study group
Azathioprine N = 32
Mycophenolate N = 16
Preswitch on azathioprine N = 231
Plus–minus values are means ± SD.
1Twenty-three of 32 patients treated with azathioprine were later switched to mycophenolate mofetil treatment.
2Glomerular filtration rate (GFR) was estimated according to the abbreviated modification of diet in renal disease study group using the following parameters: creatinine, gender, ethnicity and age (10).
3Skin type was classified according to the subjective expression of ultraviolet photosensitivity based on erythema and tanning reactivity according to Fitzpatrick where skin type-I always burns and never tans, skin type-II usually burns and tans less than average, skin type-III sometimes burns and tans about average, skin type-IV never burns and always tans more than average and skin type-V never burns and is brown (11).
55 ± 13
54 ± 14
59 ± 11
Sex, no. (%)
Estimated GFR (mL/min/1.73 m2)2
55 ± 14
41 ± 10
56 ± 13
Time since transplant (years)
15 ± 5
8 ± 5
14 ± 5
Immunosuppressant, no. (%)
Skin type (Fitzpatrick), no. (%)3
Prior cutaneous neoplasias, no. (%)
To examine the effects of AZA withdrawal on photosensitivity, 23 patients (16 males, 7 females) of the AZA group were switched to MMF. These patients had a mean AZA exposure of 14 years, and their skin types were representative of the distribution among the 32 AZA group patients (Table 1).
Skin photosensitivity in AZA and mycophenolate-treated patients
Treatment with AZA was associated with enhanced photosensitivity to UVA (Table 2). The mean (SD) UVA MED was twofold lower among patients in the AZA than in the MMF group (17 ± 10 J/cm2 in the AZA group as compared to 34 ± 17 J/cm2 in the MMF group). In a multivariate analysis, AZA was one of the significant factors accounting for the difference in the UVA photosensitivity. The differential photosensitivity was independent of skin type (Table S1). The mean UVB MED was similar for patients in the AZA and MMF groups (Table 2).
Table 2. Ultraviolet A and B photosensitivity expressed as minimal erythema dose among kidney transplant recipients, according to treatment group
Azathioprine (N = 32)
Mycophenolate (N = 16)
Mean difference (95% CI)
Plus–minus values are means ± SD.
Minimal erythema dose to UVA (J/cm2)
17 ± 10
34 ± 17
17 (7 to 27)
Minimal erythema dose to UVB (mJ/cm2)
92 ± 37
88 ± 23
−3 (−22 to 15)
The higher photosensitivity of patients receiving AZA was also apparent in the doses required for the induction of immediate erythema. These doses amounted to 16 ± 6 J/m2 and 20 ± 9 J/m2 (Mann–Whitney U-test, p = 0.02) among patient with AZA and mycophenolate, respectively. For pigmentation, the doses were 15 ± 10 J/m2 in patients with AZA and 27 ± 7 J/m2 in patients with mycophenolate (Mann–Whitney U-test, p < 0.001).
Skin photosensitivity after replacing AZA with MMF
The replacement of AZA with MMF-decreased photosensitivity to UVA in all 23 kidney transplant recipients (Figure 2). Three months after replacing AZA with MMF, SD MED for UVA increased to 25 ± 11 J/cm2 compared to 15 ± 8 J/cm2 before the switch (mean difference 10 J/cm2, 95% confidence interval (CI): 6–13 J/cm2, p < 0.001; Table 3). The photosensitivity to UVB remained unchanged. The change in medication did not affect the UVA doses required to induce immediate erythema (15 ± 5 and 17 ± 8 J/m2) and pigmentation (10 ± 5 and 11 ± 4 J/m2).
Table 3. Ultraviolet A and B photosensitivity expressed as minimal erythema dose among kidney transplant recipients at the day azathioprine (AZA) was replaced by mycophenolate mofetil (MMF) and 3 months later
Preswitch on AZA
3 months postswitch to MMF
Mean difference (95% CI)
Plus–minus values are means ± SD.
Minimal erythema dose to UVA (J/cm2)
15 ± 8
25 ± 11
Minimal erythema dose to UVB (mJ/cm2)
69 ± 27
81 ± 25
6-TG incorporation into DNA of peripheral blood mononuclear cells
DNA 6-TG is a UVA sensitizer that causes photochemical DNA damage. Using a sensitive fluorimetric assay, we measured the 6-TG DNA content in peripheral blood mononuclear cells in a subgroup of seven kidney transplant recipients who had switched from AZA to mycophenolate and for whom blood samples were available 3 months later. Before the switch, the range of 6-TG DNA values was between 50 and 130 pmol 6-TG per mg DNA (SD 107 ± 35 pmol 6-TG per mg DNA; Figure 3A). At 3 months, the DNA 6-TG content decreased in all patients. The mean 6-TG DNA value declined by almost 50% to 61 ± 22 pmol 6-TG per mg DNA (p < 0.001).
In all 23 patients who were phototested, there was a decreased UVA sensitivity. In seven patients for whom material was available, the decrease in skin UVA photosensitivity was associated with a significant reduced DNA 6-TG level (Figure 3B). However, DNA 6-TG was still detectable in all patients 3 months after the switch to mycophenolate. To investigate this persistence, additional measurements were performed in 5 of 7 patients at later times postswitch (range 7–28 months). Figure 3C shows one example (patient 6) of the time-dependent reduction in DNA 6-TG. It illustrates the consistent finding that DNA 6-TG remains detectable many months after the change to mycophenolate. Findings for all five patients are summarized in Figure 3(D). After 7 months, patient 2 retained more than 50% of the initial circulating DNA 6-TG. Strikingly, in the other patients, approximately one-fifth of the starting DNA 6-TG was still detectable 2 years (range 19–28 months) after AZA withdrawal. Long-term DNA 6-TG persistence was unrelated to substituting mycophenolate for AZA. DNA 6-TG was also detectable at 1 month, and as late as 6 months, after AZA withdrawal in two patients who had previously received the drug for inflammatory bowel disease (15 years at 150 mg daily) or eczema (6 months at 100 mg daily) and whose drug treatment had simply been discontinued (inset to Figure 3C).
To address whether this persistent DNA 6-TG was associated with a measurable residual skin photosensitivity, we compared the UVA MEDs at 3 months and at later times in four of these five patients. In three patients, the 3-month and 19–24-month values were not different, despite a measurable reduction in DNA 6-TG. In patient 7 the UVA MED had increased further (Figure 3E). This patient had the highest DNA 6-TG value at 3 months and the greatest reduction between then and the later time (19 months).
Skin expression of p53
In eight kidney transplant recipients switched from AZA to MMF, immunoreactive p53 nuclei were quantified in biopsies from sites that had received an MED for UVA. These measurements were performed immediately before and after 3 months changing to mycophenolate. In each of the eight patients, the number of p53 immunoreactive nuclei was reduced at 3 months. The average reduction was 50% (mean difference 18 p53 positively stained nuclei per HPF, 95% CI: 12–23 p53 positively stained nuclei per HPF, two-tailed paired t-test, p value < 0.001). An example is shown in Figure 4. These findings indicate that UVA causes less DNA damage when the level of DNA 6-TG has diminished, although the MED to UVA was twofold higher after the switch than those before the switch.
The switch of AZA to MMF was well tolerated. In all but one patient, MMF was taken without interruption or dose reduction. One patient did not tolerate MMF due to adverse gastrointestinal effects. Renal function assessed by estimated GFR and urinary total protein excretion as well as hematological parameters and cyclosporin trough levels were similar at all study time points (Table S2).
An awareness of the potential photosensitizing properties of drugs is particularly important in the field of organ transplantation in which prolonged pharmacological immunosuppression is associated with a high incidence of sunlight-related skin cancer (13,14). Our study firmly establishes that AZA immunosuppression significantly enhances the photosensitivity of skin UVA but not to UVB. Our study population included skin type-II and -III and a few skin type-IV, whereas previous results in UK persons indicated a similar role for AZA in a lighter-skinned population (7). It is, thus, likely that our results apply to Caucasians as a whole. We show that this enhanced photosensitivity correlates with the presence of the UVA chromophore 6-TG in patients’ DNA. Moreover, photosensitivity is associated with the formation of DNA damage and is largely, but not completely, reversed on replacing AZA with mycophenolate.
Bioactivation of AZA generates TG nucleotides that are incorporated into DNA as 6-TG (15). Because AZA administration is systemic, the DNA of all renewing cells will contain 6-TG. In this study, to avoid taking multiple skin biopsies, we used lymphocytes as a surrogate for skin DNA 6-TG measurements. We previously determined skin levels of DNA 6-TG of approximately 50 pmol 6-TG per mg DNA in a limited series of nontransplant patients taking AZA (7). It seems likely that the initial steady-state DNA 6-TG levels in the skin of our AZA group patients were in the same range of 50–130 pmol 6-TG per mg DNA.
The comparison of MEDs between AZA and MMF groups suggests that the enhanced UVA photosensitivity is not because of the immunosuppression per se but that it reflects an interaction between UVA and AZA, its metabolites or DNA 6-TG. AZA treatment increased susceptibility to immediate erythema and pigmentation as well as to delayed erythema. Only the latter was affected by the change to mycophenolate. Residual AZA or metabolites as well as the smaller patient number in the switch group may be reasons why only the key measure of photosensitivity, the delayed erythema, differed significantly. The effect on delayed erythema, therefore, most likely reflects the presence of DNA 6-TG.
The replacement of AZA by MMF was beneficial in terms of reduced skin photosensitivity. Although we did not address the question in detail, the finding of relatively low levels of DNA 6-TG in two patients who had stopped taking AZA without switching to an alternative immunosuppressant indicates that the changes we observed are likely related to AZA withdrawal rather than to its replacement by mycophenolate. Erythema is a reflection of unrepaired DNA damage (16–19), which is, in turn, linked to p53 stabilization (20). The reduced UVA-induced p53 activation at 3 months when DNA 6-TG levels had measurably diminished is consistent with reduced DNA photodamage. Measurements of p53 activation indicated that equally erythematogenic UVA doses induce less DNA damage after the switch to mycophenolate, which may reflect qualitative differences in UVA-induced DNA damage. In the absence of photosensitizing DNA 6-TG, UVA-induced DNA lesions, such as cyclobutane pyrimidine dimers or 8-oxoguanine, are efficiently removed by excision repair. In contrast, photochemical DNA damage induced by DNA 6-TG, includes various irreparable DNA lesions (15). Consequently, when DNA 6-TG levels are high, more of the potentially erythematogenic DNA damage will persist during the 24 h in which erythema develops.
The highly sensitive method we developed for quantifying 6-TG permitted the detection of measurable DNA 6-TG in AZA patients as late as 28 months after the change to mycophenolate. This novel finding reveals that hazards associated with AZA treatment persist for many months after switching to an alternative immunosuppressant. If the skin retains similar DNA 6-TG levels, it would maintain some UVA photosensitivity and would be at risk for chronic low-level DNA photodamage. Persistent or chronic DNA damage is linked to the development of cancer (21). In any tissue, including lymphocytes and skin, persistent DNA 6-TG represents a long-term risk of mutation and two independent studies have reported a novel AZA-associated mutation (22,23). In this regard, there is some evidence of an elevated risk of lymphoproliferative disorders in inflammatory bowel disease patients treated with thiopurines (24,25), although the effect is small and the evidence is somewhat conflicting (26).
In summary, switching patients from AZA to mycophenolate reverses the enhanced photosensitivity of the patients’ skin to UVA and attenuates the formation of photochemical DNA damage. However, the novel finding of long-term retention of significant DNA 6-TG indicates that the risk of therapy-related malignancy is not completely reversed upon AZA withdrawal.
We thank Karl G. Hofbauer for critical revision of the manuscript.
The authors of this manuscript have conflicts of interest to disclose as described by the American Journal of Transplantation. This study was supported by the Olga-Mayenfisch Foundation, EMDO Foundation, Hartmann-Müller-Foundation, University Medical Faculty Research Foundation, all in Zürich, Switzerland and by Hoffmann-LaRoche, Basel, Switzerland, and Cancer Research UK. G.F.L.H. received an unrestricted research grant from Hoffmann LaRoche AG, Basel, Switzerland, the manufacturer of MMF (CellCept®). Neither Hoffmann LaRoche nor any other funding body had any role in the trial design, collection, analysis and interpretation of the data or the writing of the report. The authors vouch for the fidelity of this report to the trial protocol.