In Silico Phototoxicity Prediction of Drugs and Chemicals by using Derek Nexus and QSAR Toolbox

Abstract

Phototoxicity testing is crucial for evaluating the potential harmful effects of pharmaceuticals and chemicals on human skin when exposed to sunlight. Traditional in vivo models involving mice, rats, guinea pigs, as well as in vitro assays such as the 3T3 Neutral Red Uptake phototoxicity assay and methods based on the use of reconstructed human epidermis, have been established for phototoxicity testing. While these approaches are extremely valuable, they are costly in terms of both time and resources. Consequently, in silico approaches based on the use of predictive software tools can offer more rapid and cost-effective phototoxicity screening solutions. With this goal in mind, the current study evaluated two in silico tools — Derek Nexus 6.1.0/Derek Knowledge Base 2020 1.0 (Lhasa Limited, UK) and the QSAR Toolbox (v 4.5) developed by the Organisation for Economic Co-operation and Development (OECD) — for their capacity to predict the phototoxicity of several substances from diverse classes. Derek Nexus and the QSAR Toolbox were both found to be very useful for predicting the phototoxicity of drugs and other chemicals. Derek Nexus predicted phototoxicity of the compounds, with a sensitivity of 63%, specificity of 93%, Positive Predictive Values of 90% and Negative Predictive Value of 69%, overall accuracy of 77% and balanced accuracy of 78%. The QSAR Toolbox achieved sensitivity of 73%, specificity of 85%, Positive Predictive Value of 85% and Negative Predictive Value of 74%, overall accuracy of 79% and balanced accuracy of 79%. The results show that Derek Nexus and the QSAR Toolbox can be usefully incorporated in the workflow of phototoxicity testing for pharmaceuticals and chemicals.

Keywords

Derek Nexus QSAR Toolbox phototoxicity

Introduction

Phototoxicity can be elicited by exposure to certain chemicals — including pharmaceuticals, cosmetics or food ingredients — that are photoactivated by ultraviolet (UV) A/B light.^1–3 Several classes of pharmaceuticals often cause phototoxic reactions in the skin and eyes at clinical doses,³ and perceptible adverse effects would lead to a reduction in medication compliance. In addition to the clinical issues, drug-induced phototoxicity has also halted the development of new drug entities, and the pharmaceutical industry and regulatory agencies have struggled to predict and/or avoid phototoxic liability.

Various in vivo models have used mice, rats and guinea pigs for phototoxicity testing.^4–7 In vitro assays, such as the 3T3 Neutral Red Uptake phototoxicity assay (3T3 NRU PT assay; OECD TG 432) and the phototoxicity test with reconstructed human epidermis (RhE Phototoxicity Test (RhE PT); OECD TG 498), have been introduced as alternative methods for phototoxicity testing. Since the publication of the ICH S10 guideline on the photosafety assessment of pharmaceuticals,⁸ by the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH), the interest in photosafety assessment has further increased as part of the pharmaceutical development process. In the ICH S10 guideline, absorption of sunlight (290–700 nm), generation of reactive species (such as reactive oxygen species (ROS)), and distribution to light-exposed tissues (including skin and eyes), have been described as key characteristics of phototoxic chemicals.⁸

Given the considerable expense and time requirements associated with traditional in vivo and in vitro assays, we sought a more efficient and cost-effective approach. Our plan involved conducting an in silico evaluation of multiple substances from diverse classes by using two well-established knowledge-based (Quantitative) Structure–Activity Relationship ((Q)SAR) systems — namely, the Organisation for Economic Co-operation and Development (OECD) QSAR Toolbox and Derek Nexus. This innovative strategy aimed to expedite the phototoxicity screening of chemicals and drugs, while also providing an economical solution for assessing potential risks.

Materials and methods

A set of 57 non-proprietary compounds were analysed by using the in silico tools Derek Nexus 6.1.0/Derek Knowledge Base 2020 1.0 (Lhasa Limited, UK) and the QSAR Toolbox v4.5 (OECD, France). The compounds chosen represented a diverse range of substances, including phototoxic/non-phototoxic chemicals and drugs (30 phototoxic and 27 non-phototoxic). To supplement the analysis, in vitro 3T3 NRU PT assay and RhE PT data for the selected substances were gathered, as well as animal and human data on phototoxicity. These compounds have relatively broad use in the chemical, cosmetics, pharmaceutical, and medical device industries.

Compounds were entered into Derek using SMILES strings. From the Derek Prediction Set Up, ‘phototoxicity’ prediction only was selected from the miscellaneous endpoints menu. When evaluating the results from Derek Nexus, various degrees of phototoxicity predictions were considered — such as ‘probable’, ‘plausible’ or ‘equivocal’ — as positive indicators for phototoxicity. Conversely, a negative result was assigned in cases where the query compounds did not match any phototoxicity alerts.

The OECD Toolbox was used in Classic User Interface mode. Compounds were entered into the Toolbox using the CAS number. For each compound, data available in the Toolbox from the photosensitivity database, was sought (under the category of Human Health Hazards, photoinduced toxicity). Information on this database (as stated in the OECD Toolbox) reports that this database contains photosensitisation data for chemicals evaluated by weight-of-evidence. The weight-of-evidence approach involves using a combination of information from several independent sources to give sufficient evidence to fulfil an information requirement. According to the information provided on the OECD Toolbox model, the weight-of-evidence evaluation is based on the following: “(1) Negative in phototoxicity test (in vivo data, in vitro 3T3 Neutral Red Uptake phototoxicity test; (2) Positive in in vivo photosensitization test (Adjuvant-Strip test); (3) Photosensitization information is available for structurally similar compounds”. For the QSAR Toolbox, we focused solely on positive and negative results in relation to phototoxicity (i.e. when gathering data under the category ‘photoinduced toxicity’, results for photoallergy and photoirritation were not considered). Where phototoxicity data were not available within the Toolbox, ‘data gap filling’ was used, by executing the existing QSAR model for photoinduced toxicity from the Toolbox.⁹ In this analysis, to enable comparison with results from Derek Nexus (where a no alert result can be considered as out of domain), results from the Toolbox were recorded as negative, if the prediction from the QSAR Toolbox model was negative or the compound was outside of the applicability domain of the QSAR Toolbox model. Further details of the model employed within the QSAR Toolbox are given by Ringeissen et al.¹⁰

Results

Our study yielded significant insights into the predictive performance of Derek Nexus and of the QSAR Toolbox for phototoxicity assessment. A summary of our results can be found in Table 1, where we compiled the phototoxicity alerts provided by Derek Nexus and the QSAR Toolbox. Based on the compiled information, the compounds were classified as True Positive (TP), False Positive (FP), True Negative (TN) and False Negative (FN), as presented in Figure 1.

Figure 1.

Phototoxicity prediction by using Derek Nexus and the QSAR Toolbox. TP = true positive; FN = false negative; FP = false positive; TN = true negative; +ve = positive, i.e effect observed; –ve = negative, i.e no effect observed; PPV = Positive predicted value; NPV = Negative predicted value.

Table 1.

Phototoxicity data including in silico predictions, in vitro and in vivo data for the various substances tested.

Chemical	CAS No.	Result			Overall call for phototoxicity^a	Derek outcome	Derek call for phototoxicity^b	Derek predictivity for phototoxicity	Derek toxicity alerts	OECD Toolbox call for phototoxicity^c	OECD Toolbox predictivity for phototoxicity
Chemical	CAS No.	3T3 NRU PT assay	3-D tissue models	In vivo	Overall call for phototoxicity^a	Derek outcome	Derek call for phototoxicity^b	Derek predictivity for phototoxicity	Derek toxicity alerts	OECD Toolbox call for phototoxicity^c	OECD Toolbox predictivity for phototoxicity
Acridine	260-94-6	PT⁵	PT²³	PT²⁴	+ve	No alert	–ve	False negative	No alert	+ve	True positive
Amiodarone HCL	19774-82-4	PT²⁵	PT²⁶	PT²⁵	+ve	No alert	–ve	False negative	No alert	–ve	False negative
Chlorpromazine HCL	69-09-0	PT²⁷	PT²⁸	PT⁷	+ve	PT probable	+ve	True positive	Phenothiazine or analogue	+ve	True positive
Doxycycline HCL	10592-13-9	PT²⁹	—	PT^30,31	+ve	PT plausible	+ve	True positive	Tetracycline	–ve	False negative
Fenofibrate	49562-28-9	PT²⁷	PT²⁷	PT²⁹	+ve	No alert	–ve	False negative	No alert	–ve	False negative
Furosemide	54-31-9	NPT²⁷	PT²⁷	PT³²	+ve	PT probable	+ve	True positive	Aryl sulphonamide	+ve	True positive
Ketoprofen	22071-15-4	PT²⁹	PT²⁷	PT²⁹	+ve	PT plausible	+ve	True positive	Arylacetic or 2-arylpropionic acid	+ve	True positive
8-Methoxypsoralen	298-81-7	PT⁶	PT²³	PT⁶	+ve	No alert	–ve	False negative	No alert	+ve	True positive
Nalidixic acid	389-08-2	—	—	PT⁷	+ve	PT plausible	+ve	True positive	Quinolone-3-carboxylic acid or naphthyridine analogue	+ve	True positive
Norfloxacin	70458-96-7	PT³³	—	PT³⁴	+ve	PT plausible	+ve	True positive	Quinolone-3-carboxylic acid or naphthyridine analogue	+ve	True positive
Ofloxacin	82419-36-1	PT²⁵	PT³⁵	PT²⁵	+ve	PT probable	+ve	True positive	Quinolone-3-carboxylic acid or naphthyridine analogue	+ve	True positive
Promethazine	60-87-7	PT⁵	PT²⁷	PT²⁵	+ve	PT probable	+ve	True positive	Phenothiazine or analogue	+ve	True positive
Tetracycline	60-54-8	NPT⁵	PT³⁶	PT³⁷	+ve	PT plausible	+ve	True positive	Tetracycline	+ve	True positive
Lomefloxacin	98079-51-7	PT⁶	PT²⁷	PT⁶	+ve	PT probable	+ve	True positive	Quinolone-3-carboxylic acid or naphthyridine analogue	+ve	True positive
Demeclocycline hydrochloride	64-73-3	PT²⁷	PT²⁷	PT²⁷	+ve	PT plausible	+ve	True positive	Tetracycline	–ve	False negative
Tiaprofenic acid	33005-95-7	PT²⁷	PT²⁷	PT²⁷	+ve	PT probable	+ve	True positive	Arylacetic or 2-arylpropionic acid	–ve	False negative
Phenothiazine	92-84-2	PT²⁷	PT²⁷	PT³⁸	+ve	PT plausible	+ve	True positive	Phenothiazine or analogue	+ve	True positive
Angelicin (isopsoralen)	523-50-2	—	Equivocal²⁷	PT²⁷	+ve	No alert	–ve	False negative	No alert	+ve	True positive
Quinine HCL dihydrate	6119-47-7	PT³⁹	—	PT³⁸	+ve	PT plausible	+ve	True positive	4-Quinolinyl methanol	+ve	True positive
Sparofloxacin	110871-86-8	PT⁶	—	PT⁶	+ve	PT plausible	+ve	True positive	Quinolone-3-carboxylic acid or naphthyridine analogue	+ve	True positive
Pirfenidone	53179-13-8	PT⁶	—	PT ⁶	+ve	No alert	–ve	False negative	No alert	+ve	True positive
Vemurafenib	918504-65-1	PT⁴⁰	—	PT⁴⁰	+ve	No alert	–ve	False negative	No alert	+ve	True positive
Demeclocycline	127-33-3	—	—	PT³⁰	+ve	PT plausible	+ve	True positive	Tetracycline	–ve	False negative
Protoporphyrin IX	553-12-8	PT⁵	—	PT⁴¹	+ve	No alert	–ve	False negative	No alert	–ve	False negative
Enoxacin	74011-58-8	PT²⁹	—	PT²⁹	+ve	PT plausible	+ve	True positive	Quinolone-3-carboxylic acid or naphthyridine analogue	+ve	True positive
Aspirin	50-78-2	NPT⁴²	—	NPT³	–ve	NPT	–ve	True negative	No alert	–ve	True negative
Benzocaine	94-09-7	NPT⁴²	—	NPT³	–ve	NPT	–ve	True negative	No alert	–ve	True negative
Erythromycin	114-07-8	NPT⁴²	—	NPT³	–ve	NPT	–ve	True negative	No alert	–ve	True negative
Phenytoin	57-41-0	NPT⁴²	—	NPT³	–ve	NPT	–ve	True negative	No alert	–ve	True negative
Penicillin G	61-33-6	NPT³⁷	NPT²⁷	NPT²⁷	–ve	No alert	–ve	True negative	No alert	–ve	True negative
6-Methylcoumarin	92-48-8	PT⁴³	NPT³⁶	NPT⁴⁴	–ve	No alert	–ve	True negative	No alert	+ve	False positive
Piroxicam	36322-90-4	NPT³⁷	—	NPT³⁷	–ve	PT plausible	+ve	False positive	Aryl sulphonamide	–ve	True negative
Anthracene	120-12-7	PT⁴⁵	PT²³	PT⁴⁶	+ve	No alert	–ve	False negative	No alert	+ve	True positive
Bithionol	97-18-7	PT⁴⁷	—	PT²⁵	+ve	PT probable	+ve	True positive	Halogenated Phenol	+ve	True positive
Neutral red	553-24-2	PT⁴⁷	PT³⁶	PT²⁵	+ve	No alert	–ve	False negative	No alert	+ve	True positive
Bengal rose	11121-48-5	PT²⁵	PT²⁶	PT²³	+ve	PT plausible	+ve	True positive	Fluorescein or derivative	–ve	False negative
5-Methoxypsoralen	484-20-8	PT⁴⁷	PT²⁷	PT²⁵	+ve	No alert	–ve	False negative	No alert	+ve	True positive
4-Methyl benzylidene camphor	36861-47-9	NPT⁴⁸	—	NPT⁴⁹	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Benzophenone-3	131-57-7	NPT⁴⁸	NPT³⁶	NPT⁵⁰	–ve	No alert	–ve	True negative	No alert	–ve	True negative
l-Histidine	71-00-1	NPT⁴⁷	NPT²⁷	NPT²⁷	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Mexoryl SX	92761-26-7	—	—	NPT⁵¹	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Octyl methoxy cinnamate	5466-77-3	NPT⁴⁸	NPT²³	NPT²³	–ve	No alert	–ve	True negative	No alert	+ve	False positive
Octyl salicylate	6969-49-9	NPT⁴⁷	NPT⁵²	NPT⁵³	–ve	No alert	–ve	True negative	No alert	–ve	True negative
PARA-aminobenzoic acid (PABA)	150-13-0	NPT²⁵	NPT³⁶	NPT²⁵	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Sodium lauryl sulphate or SDS	151-21-3	NPT⁴⁵	NPT²⁷	NPT²⁷	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Ethyl vanillin	121-32-4	PT⁵⁴	NPT⁵⁴	NPT⁵⁴	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Titanium dioxide	13463-67-7	PT⁵⁵	NPT⁵⁶	NPT⁵⁷	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Vanillin isobutyrate	20665-85-4	PT⁵⁴	NPT⁵⁴	NPT⁵⁴	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Cinnamaldehyde	104-55-2	PT⁵⁸	NPT⁵⁸	NPT⁵⁸	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Cadmium selenide	1306-24-7	NPT⁵⁹	NPT⁵⁹	—	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Mercuric sulphide	1344-48-5	NPT⁵⁹	NPT⁵⁹	—	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Chromium oxide	12018-34-7	NPT⁵⁹	NPT⁵⁹	—	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Cobalt aluminate	1333-88-6	NPT⁵⁹	NPT⁵⁹	—	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Coumarin	91-64-5	—	NPT²⁷	NPT²⁷	–ve	No alert	–ve	True negative	No alert	+ve	False positive
Avobenzone	70356-09-1	PT⁴⁸	NPT²³	NPT⁵⁰	–ve	No alert	–ve	True negative	No alert	–ve	True negative
Hexachlorophene	70-30-4	NPT⁶⁰	—	NPT⁶¹	–ve	PT probable	+ve	False positive	Halogenated phenol	–ve	True negative
Musk ambrette	83-66-9	PT⁴⁷	NPT⁶²	NPT²⁵	–ve	No alert	–ve	True negative	No alert	+ve	False positive

^aThis column refers to the collective results obtained with the 3T3 NRU PT assay, 3-D tissue models and historical in vivo data.

^bCall for phototoxicity, based on the outcome of the Derek in silico model.

^cCall for phototoxicity, based on the outcome from the OECD Toolbox, where +ve = a positive result as recorded in the photoxicity database within the Toolbox, or a positive prediction based on the phototoxicity QSAR model available in the Toolbox; –ve = a negative result as recorded in the photoxicity database within the Toolbox or out of domain of the QSAR model (please refer to the main text).

PT = phototoxicity; NPT = no phototoxicity; +ve = positive, i.e effect observed; –ve = negative, i.e. no effect observed.

Derek Nexus showed a sensitivity of 63%, specificity of 93%, Positive Predictive Value (PPV) of 90%, Negative Predictive Value (NPV) of 69%, overall accuracy of 77% and balanced accuracy of 78%. Similarly, the QSAR Toolbox showed a sensitivity of 73%, specificity of 85%, PPV of 85%, NPV of 74%, overall accuracy of 79% and balanced accuracy of 79%. The formulae employed for calculating these predictive parameters were as follows:

Sensitivity = [True Positive / (True Positive + False Negative)] \times 100

Specificity = [True Negative / (True Negative + False Positive)] \times 100

Positive Predictive Value (PPV) = [True Positive / (True Positive + False Positive)] \times 100

Negative Predictive Value (NPV) = [True Negative / (True Negative + False Negative)] \times 100

Overall Accuracy = [True Positive + True Negative / (True Positive + True Negative + False Positive + False Negative)] \times 100

Balanced accuracy = Sensitivity + Specificity / 2

It is essential to acknowledge that both Derek Nexus and the QSAR Toolbox predictions are applicable only to substances with known chemical structures. In contrast, predicting the phototoxic potential of mixtures, formulations and oils (such as plant extracts and cosmetics) with unknown compositions, remains difficult using in silico methods.

Our findings highlight the potential of incorporating QSARs into the workflow for assessing phototoxicity in pharmaceuticals, particularly within the context of the ICH S10 guidelines. Additionally, QSARs can be valuable for the initial phototoxicity evaluation of chemicals.

Discussion

Exogenous compounds that reach the skin after topical application or after oral exposure, such as pharmaceuticals and personal care products, can be activated by solar radiation and may contribute to adverse effects in the skin, such as photoirritation, photo-induced ageing or photocarcinogenicity.^2,11,12 As a result, photosafety testing has become a mandatory regulatory requirement for consumer products that absorb light in the range of 290–700 nm and with (relevant) partitioning in the skin or eyes. Guidance documents outlining the strategy and approaches for the photosafety evaluation of pharmaceutical products are provided by both the European Medicines Agency (EMA) and the US Food and Drug Administration (FDA) Center for Drug Evaluation and Research (CDER).^13,14 In addition, an OECD guidance document on phototoxicity testing of chemicals is in preparation.

Testing of pharmaceutical compounds and personal care products may include evaluating acute phototoxicity (photoirritation), photoallergenicity, photogenotoxicity and photocarcinogenicity.¹⁵ A limited number of knowledge-based expert systems are currently available for predicting photo-induced toxicity. This is not surprising, since establishing the phototoxic potential of chemical entities is a complex task, due to the various mechanisms involved in photoirritation, photogenotoxicity or photoallergenicity reactions.¹⁶ Derek Nexus is one of the tools containing structural alerts for photo-induced toxicity.¹⁷ However, Ringeissen and coworkers found that Derek was not successful at identifying the phototoxicity of compounds tested with the 3T3 NRU PT assay.¹⁰

The QSAR paradigm is endorsed by one of the most widely used in silico tools in regulatory settings: the OECD QSAR Toolbox. The OECD QSAR Toolbox is a freely available in silico system that comprises several databases, and can be used to provide information on toxicological, ecotoxicological and physicochemical properties of substances. The Toolbox enables users to group chemicals within categories based on mechanistic (i.e. a similar mode of action) and structural (e.g. same organic functional group) similarities, and then predict toxicological properties.¹⁰ Whilst grouping and read-across are potential uses of the Toolbox, within the present study, predictions were based on the application of a QSAR model, rather than a read-across approach. The main local (Q)SAR models (i.e. those derived from a small set of similar chemical compounds) developed so far to predict the phototoxic effects of some specific classes of chemicals apply to fluoroquinolones,¹⁸ quinine derivatives,¹⁹ phenyl and benzoyl pyrroles,²⁰ tricyclic thiophenes,²¹ polycyclic aromatic hydrocarbons (PAH), and a broad set of pharmaceuticals.²² The predictions obtained from the Toolbox rely upon a database containing 73 photosensitising chemicals (antibacterial agents, NSAIDs and cosmetic materials), and are obtained by weight-of-evidence and the supporting data.

In the current study, we obtained a balanced accuracy of 78% by using Derek Nexus and 79% with the QSAR Toolbox. Our analysis revealed that Derek Nexus (v6.1.0) and QSAR Toolbox (v4.5) are effective in predicting phototoxicity. We conducted an in silico assessment of 57 diverse drugs and chemicals, encompassing phototoxic drugs, non-phototoxic drugs, phototoxic chemicals and non-phototoxic chemicals. To the best of our knowledge, this study represents the most comprehensive attempt to date to utilise computational toxicology for phototoxicity prediction.

Conclusions

In conclusion, our study demonstrates the value of Derek Nexus and QSAR Toolbox in accurately predicting the phototoxicity of drugs and chemicals. Derek achieved sensitivity of 63%, specificity of 93%, a Positive Predictive Value (PPV) of 90%, a Negative Predictive Value (NPV) of 69%, an overall accuracy of 77% and a balanced accuracy of 78%. Similarly, the QSAR Toolbox achieved commendable predictivity, showing a sensitivity of 73%, specificity of 85%, PPV of 85%, NPV of 74%, overall accuracy of 79% and balanced accuracy of 79%. Notably, to the best of our knowledge, Derek provided predictive alerts for phototoxicity that were unavailable through any other in silico prediction tool, adding to its unique value.

However, for robust validation, further assessment of Derek Nexus and the QSAR Toolbox with a broader range of test compounds is warranted in order to address the broader spectrum of applicability domains, e.g. medical devices, pesticides, food and animal feed.

Despite this need for further assessment, the current results strongly support the incorporation of both Derek Nexus and of the QSAR Toolbox into the workflow of phototoxicity testing for pharmaceuticals, some raw materials used in cosmetics and medical devices, and other chemicals. The predictive capabilities exhibited by these tools offer a valuable means of streamlining phototoxicity evaluation, contributing to enhanced safety assessments in drug and chemical development while avoiding animal use.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was partly supported by APVV Research Grant APVV-19-0591: In vitro biocompatibility assessment of medical devices (MD) and innovative bio-materials for MD.

References

Moore

. Mechanisms of photosensitization by phototoxic drugs. Mutat Res 1998; 422: 165–173.

Moore

. Drug-induced cutaneous photosensitivity: Incidence, mechanism, prevention and management. Drug Saf 2002; 25: 345–372.

Onoue

Seto

Gandy

, et al. Drug-induced phototoxicity; an early in vitro identification of phototoxic potential of new drug entities in drug discovery and development. Curr Drug Saf 2009; 4: 123–136.

Blotz

Michel

Moysan

, et al. Analyses of cutaneous fluoroquinolones photoreactivity using the integrated model for the differentiation of skin reactions. J Photochem Photobiol B, Biol B 2000; 58: 46–53.

Neumann

Blotz

Wasinska-Kempka

, et al. Evaluation of phototoxic and photoallergic potentials of 13 compounds by different in vitro and in vivo methods. J Photochem Photobio B Biol 2005; 79: 25–34.

Adachi

Satou

, et al. Assessment of 8-methosypsoralen, lomefloxacin, sparfloxacin, and pirfenidone phototoxicity in Long-Evans rats. Int J Toxicol 2015; 34: 16–23.

Kim

Youn

J-S

, et al. Evaluation of the skin phototoxicity of systemically administered pharmaceuticals in Sprague-Dawley rats. Toxicol Res 2021; 37: 435–442.

ICH. ICH Guideline S10 : Guidance on Photosafety Evaluation of Pharmaceuticals, https://database.ich.org/sites/default/files/S10_Guideline.pdf (2013, accessed 7 March 2024).

OECD . OECD QSAR Toolbox v4.5, https://qsartoolbox.org/ (accessed 7 March 2024).

10.

Ringeissen

Marrot

Notea

, et al. Development of a mechanistic SAR model for the detection of phototoxic chemicals and use in an integrated testing strategy. Toxicol In Vitro 2011; 25: 324–334.

11.

Lankerani

Baron

. Photosensitivity to exogenous agents. J Cutan Med 2004; 8: 424–431.

12.

Marrot

Meunier

. Skin DNA photodamage and its biological consequences. J Am Acad Dermatol 2008; 58: S139–S148.

13.

EMEA/CPMP 2002 . Note for guidance on photosafety testing, CPMP/SWP/398/01. London: The European Agency for the Evaluation of Medicinal Products, 2002, 8 pp.

14.

FDA/CDER . Guidance for industry: Photosafety testing, https://www.govinfo.gov/content/pkg/GOVPUB-HE20-PURL-LPS120121/pdf/GOVPUB-HE20-PURL-LPS120121.pdf (2003, accessed 8 March 2024).

15.

Maurer

. Phototoxicity testing — in vivo and in vitro. Food Chem Toxicol. 1987; 25: 407–414.

16.

Onoue

Tsuda

. Analytical studies on the prediction of photosensitive/phototoxic potential of pharmaceutical substances. Pharm Res 2006; 23: 156–164.

17.

Marchant

Briggs

Long

. In silico tools for sharing data and knowledge on toxicity and metabolism: Derek for windows, meteor, and vitic. Toxicol Mech Meth. 2008; 18: 177–187.

18.

Dimitrov

Diderich

Sobanski

, et al. QSAR Toolbox — workflow and major functionalities. SAR QSAR Environ Res 2016; 27: 203–219.

19.

Hayashi

. New findings on the structure-phototoxicity relationship and photostability of fluoroquinolones. Yakugaku Zasshi 2005; 125: 255–261.

20.

Waddell

Carter

Arnason

, et al. Phototoxicity of quinine derivatives. Structure–activity trends. Fitoterapia (Milano) 1997; 68: 381–383.

21.

Laatsch

Renneberg

Hanefeld

, et al. Structure–activity relationship of phenyl-and benzoylpyrroles. Chem Pharm Bull 1995; 43: 537–546.

22.

Marles

Hudson

Graham

, et al. Structure–activity studies of photoactivated antiviral and cytotoxic tricyclic thiophenes. Photochem Photobiol 1992; 56: 479–487.

23.

Schmidt

Evers

Amberg

, et al. Phototoxicity — from molecular descriptors to classification models. J Cheminformatics 2011; 3: O1.

24.

Jones

Lovell

King

, et al. In vitro testing for phototoxic potential using the EpiDerm 3-D reconstructed human skin model. Toxicology Methods 2001; 11: 1–9.

25.

Yonezawa

Miyashita

Ashizawa

, et al. Evaluation of a general toxicity study incorporating phototoxicity assessments in Sprague-Dawley rats. J Toxicol Sci 2017; 42: 145–157.

26.

Spielmann

Balls

Dupuis

, et al. The international EU/COLIPA in vitro phototoxicity validation study: Results of phase II (blind trial). Part 1: The 3T3 NRU phototoxicity test. Toxicol In Vitro. 1998; 12: 305–327.

27.

Liebsch

Döring

Donelly

, et al. Application of the human dermal model Skin2 ZK 1350 to phototoxicity and skin corrosivity testing. Toxicol In Vitro 1995; 9: 557–562.

28.

Lelièvre

Justine

Christiaens

, et al. The EpiSkin phototoxicity assay (EPA): Development of an in vitro tiered strategy using 17 reference chemicals to predict phototoxic potency. Toxicol In Vitro. 2007; 21: 977–995.

29.

Zlabiene

Baranauskaite

Kopustinskiene

, et al. In vitro and clinical safety assessment of the multiple W/O/W emulsion based on the active ingredients from Rosmarinus officinalis L., Avena sativa L. and Linum usitatissimum L. Pharmaceutics. 2021; 13: 732.

30.

Schümann

Boudon

Ulrich

, et al. Integrated preclinical photosafety testing strategy for systemically applied pharmaceuticals. Toxicol Sci 2014; 139: 245–256.

31.

Kloss

McAfee

Mandel

, et al. An in vivo method for the assessment of drug-induced phototoxicity using the mouse ear. Toxicology Methods. 1991; 1: 252–262.

32.

Lee

Lim

, et al. Phototoxicity evaluation of pharmaceutical substances with a reactive oxygen species assay using ultraviolet A. Toxicol Res 2017; 33: 43–48.

33.

Yonezawa

Ohsumi

Miyashita

, et al. Evaluation of skin phototoxicity study using SD rats by transdermal and oral administration. J Toxicol Sci 2015; 40: 667–683.

34.

Radko

Minta

Stypuła-Trębas

. Influence of fluoroquinolones on viability of Balb/c 3T3 and HepG2 cells. J Vet Res 2013; 57: 599–606.

35.

Horio

Miyauchi

Asada

, et al. Phototoxicity and photoallergenicity of quinolones in guinea pigs. J Dermatol Sci 1994; 7: 130–135.

36.

Portes

Pygmalion

Popovic

, et al. Use of human reconstituted epidermis Episkin® for assessment of weak phototoxic potential of chemical compounds. Photodermatol Photoimmunol Photomed 2002; 18: 96–102.

37.

Bernard

Barrault

Deguercy

, et al. Development of a highly sensitive in vitro phototoxicity assay using the SkinEthic^TM reconstructed human epidermis. Cell Biol Toxicol. 2000; 16: 391–400.

38.

Spielmann

Balls

Brand

, et al. EEC/COLIPA project on in vitro phototoxicity testing: First results obtained with a Balb/c 3T3 cell phototoxicity assay. Toxicol In Vitro. 1994; 8: 793–796.

39.

Onoue

Kato

Inoue

, et al. Photosafety screening of phenothiazine derivatives with combined use of photochemical and cassette-dosing pharmacokinetic data. Toxicol Sci 2014; 137: 469–477.

40.

Onoue

Suzuki

Kato

, et al. Non-animal photosafety assessment approaches for cosmetics based on the photochemical and photobiochemical properties. Toxicol In Vitro 2013; 27: 2316–2324.

41.

Boudon

Plappert-Helbig

Odermatt

, et al. Characterization of vemurafenib phototoxicity in a mouse model. Toxicol Sci 2014; 137: 259–267.

42.

Gerscher

Connelly

Beijersbergen Van Henegouwen

, et al. A quantitative assessment of protoporphyrin IX metabolism and phototoxicity in human skin following dose‐controlled delivery of the prodrugs 5‐aminolaevulinic acid and 5‐aminolaevulinic acid‐n‐pentylester. Br J Dermatol 2001; 144: 983–990.

43.

Onoue

Ochi

Gandy

, et al. High-throughput screening system for identifying phototoxic potential of drug candidates based on derivatives of reactive oxygen metabolites. Pharm Res 2010; 27: 1610–1619.

44.

Toyoda

Sugiyama

Furihata

, et al. Development of a modified 3T3 Neutral Red Uptake Phototoxicity Test protocol for evaluation of poorly water-soluble substances. J Toxicol 2017; 42: 569–577.

45.

Seto

Ohtake

Sato

, et al. Phototoxic risk assessment of dermally-applied chemicals with structural variety based on photoreactivity and skin deposition. Regul Toxicol Pharmacol 2020; 113: 104619.

46.

OECD . Test No. 432: In Vitro 3T3 NRU Phototoxicity Test, OECD Guidelines for the Testing of Chemicals, Section 4. Paris: OECD Publishing, 2019, 20pp.

47.

Walter

DeQuoy

. Anthracene with near ultraviolet light inhibiting epidermal proliferation. Arch Dermatol 1978; 114: 1463–1465.

48.

Haranosono

Kurata

Sakaki

. Establishment of an in silico phototoxicity prediction method by combining descriptors related to photo-absorption and photo-reaction. J Toxicol Sci. 2014; 39: 655–664.

49.

Gaspar

Tharmann

Campos

, et al. Skin phototoxicity of cosmetic formulations containing photounstable and photostable UV-filters and vitamin A palmitate. Toxicol In Vitro 2013; 27: 418–425.

50.

Scientific Committee on Consumer Safety . SCCS scientific opinion on 4-methylbenzylidene camphor (4-MBC), preliminary version of 22 December, final version of 29 April 2022, SCCS/1640/21.

51.

Benevenuto

Guerra

Gaspar

. Combination of retinyl palmitate and UV-filters: Phototoxic risk assessment based on photostability and in vitro and in vivo phototoxicity assays. Eur J Pharm Sci. 2015; 68: 127–136.

52.

Pavlou

Siamidi

Vlachou

, et al. UV filters and their distribution on the skin through safe. Non-penetrating vehicles. J Cosm Sci 2021; 72: 298–324.

53.

Price

Arthur

Weiger

, et al. Modification in procedure of the Reconstructed Human EpiDermis (RhE) phototoxicity assay. Institute for In Vitro Sciences, Gaithersburg, MD, USA. 14 April 2022, Abstract/Poster 4208/P138.

54.

OECD . Test No. 498: In vitro phototoxicity — Reconstructed human epidermis phototoxicity test method, OECD Guidelines for the Testing of Chemicals, Section 4. Paris: OECD Publishing, 2023, 28pp.

55.

Ritacco

Hilberer

Lavelle

, et al. Use of alternative test methods in a tiered testing approach to address photoirritation potential of fragrance materials. Regul Toxicol Pharmacol 2022; 129: 105098.

56.

Jin

Kim

, et al. In vitro and in vivo toxicological evaluation of transition metal-doped titanium dioxide nanoparticles: Nickel and platinum. Mater Sci Eng C. 2020; 115: 110843.

57.

Líšková

Letašiová

Jantová

, et al. Evaluation of phototoxic and cytotoxic potential of TiO2 nanosheets in a 3D reconstructed human skin model. ALTEX 2020; 37: 441–450.

58.

Scientific Committee on Consumer Safety . Opinion on titanium dioxide (nano form), COLIPA no. S75, 22 July 2013, revision of 22 April 2014.

59.

Api

Belmonte

Belsito

, et al. RIFM fragrance ingredient safety assessment, cinnamaldehyde, CAS Registry Number 104-55-2. Food Chem Toxicol 2019; 134: 110710.

60.

Kim

Seo

Choi

, et al. Evaluation of phototoxicity of tattoo pigments using the 3T3 Neutral Red Uptake phototoxicity test and a 3D human reconstructed skin model. Toxicol In Vitro 2020; 65: 104813.

61.

Maciel

Moreira

Carmo

, et al. Implementation of an in vitro methodology for phototoxicity evaluation in a human keratinocyte cell line. Toxicol In Vitro. 2019; 61: 104618.

62.

Harber

Targovnik

Baer

. Studies on contact photosensitivity to hexachlorophene and trichlorocarbanilide in guinea pigs and man. J Invest Dermatol 1968; 51: 373–377.

63.

Liebsch

Barrabas

Traue

, et al. Development of a new in vitro test for dermal phototoxicity using a model of reconstituted human epidermis. ALTEX. 1997; 14: 165–174.