Lessons Learned: Transfer of the High-Definition Circulating Tumor Cell Assay Platform to Development as a Commercialized Clinical Assay Platform.

Planning and transfer of a new technology platform developed in an academic setting to a start-up company for medical diagnostic product development may appear daunting and costly in terms of complexity, time, and resources. In this review we outline the key steps taken and lessons learned when a technology platform developed in an academic setting was transferred to a start-up company for medical diagnostic product development in the interest of elucidating development toolkits for academic groups and small start-up companies starting on the path to commercialization and regulatory approval.

Progress in cancer chemoprevention: perspectives on agent selection and short-term clinical intervention trials.

The basic cancer-related chemical and biological sciences, pathology, and epidemiology have contributed to the understanding that antimutagenesis and antiproliferation are the important general mechanisms of chemoprevention and to the development of antimutagenic and anti-proliferative agents as potential chemopreventive drugs. These disciplines have also provided the biochemical and histopathological bases for identifying intermediate biomarkers that can be used as surrogate end points for cancer incidence in clinical chemoprevention trials and for selecting cohorts for these trials. Particularly important as histological biomarkers of cancer are the cytonuclear morphological and densitometric changes that define intraepithelial neoplasia (IEN). IEN changes are on the causal pathway to cancer. They may serve as target lesions in Phase II chemoprevention trials and as standards against which other earlier cellular and molecular biomarkers can be evaluated. Strategies for the clinical evaluation of chemopreventive agents have been defined for seven targets--colorectal, prostate, lung, breast, bladder, oral, and cervical cancers. Cohorts have been identified for short-term Phase II trials that investigate the effects of chemopreventive agents on IEN and on earlier biomarkers. Patients with adenomas serve as a cohort for trials in colon. One cohort for Phase II trials in prostate is patients with early stage cancers scheduled for prostatectomy; another is patients with prostatic intraepithelial neoplasia (without prostatic carcinoma). Patients treated for lung cancer are at high risk for bronchial dysplasia and second cancers; such patients are a cohort for Phase II trials in lung cancer. Presurgical breast cancer patients and patients with ductal or lobular carcinoma in situ are cohorts for studies in breast. Patients with superficial bladder cancers (Ta/T1 with or without carcinoma in situ) are cohorts for studies of chemoprevention in bladder, and patients with dysplastic oral leukoplakia are evaluated for chemoprevention of oral cancers. Cervical intraepithelial neoplasia is a prototype IEN, and patients with cervical intraepithelial neoplasia are a cohort for studies of cervical cancer.

A systematic search for structure-activity relationships of skin contact sensitizers: methodology.

A computerized resource for the systematic evaluation of the structure-activity relationships and other aspects of contact allergens is described. This resource consists of a data base of results of contact dermatitis tests and a structural classification scheme for contact allergens that is called a Structure-Activity (S/A) Tree. The data base now contains approximately 2200 test results extracted from the journal Contact Dermatitis (1975-1982) and is continually being expanded. The S/A Tree is being developed to provide an index to structure-activity relationships of contact allergens; 63 structural groups are currently indexed. Analyses of benzoquinones and gallic acid esters are presented as examples of the potential application of this resource to such problems as the identification of potential cross-reactants, appropriate test concentrations and vehicles, and the reliability of available test results.

Chemopreventive drug development: perspectives and progress.

Chemoprevention drug development has the goal of identifying safe and effective chemopreventive agents for clinical use. Several distinctive strategies are pursued in developing chemopreventive agents: (a) identifying and validating predysplastic and early dysplastic lesions that can be used instead of cancers as endpoints for measuring chemopreventive activity; (b) identifying and testing candidate agents based on considerations of mechanisms of action; (c) evaluating combinations of agents with potential for maximizing efficacy and minimizing toxicity; and (d) applying a systematic methodology for identifying and ranking candidate agents at each stage of development to ensure discovery of the best agents and most effective use of available resources. This article discusses 22 drugs and three drug combinations which have reached an advanced stage of development as chemopreventive agents. The first generation of drugs are the most advanced, now being in Phase II and Phase III clinical trials. These drugs include several retinoids [vitamin A, 13-cis-retinoic acid, all-trans-N-(4-hydroxyphenyl)retinamide], calcium, beta-carotene, tamoxifen, and finasteride. The second generation drugs are those in Phase I clinical trials. From most to least advanced, these drugs are 2-difluoromethylornithine, sulindac, piroxicam, oltipraz, N-acetyl-I-cysteine, aspirin, ibuprofen, carbenoxolone, 18 beta-glycyrrhetinic acid, and the combination of 2-difluoromethylornithine with piroxicam. The third generation includes agents with significant evidence of chemopreventive activity in animal models. These agents are now in preclinical toxicity testing. They are S-allyl-I-cysteine, phenhexyl isothiocyanate, curcumin, ellagic acid, fumaric acid, fluasterone, and the combinations of all-trans-N-(4-hydroxyphenyl)retinamide with oltipraz and all-trans-N-(4-hydroxyphenyl) retinamide with tamoxifen.

Epidermal growth factor receptor tyrosine kinase inhibitors as potential cancer chemopreventives.

Among the most important targets for chemopreventive intervention and drug development are deregulated signal transduction pathways, and protein tyrosine kinases are key components of these pathways. Loss of tyrosine kinase regulatory mechanisms has been implicated in neoplastic growth; indeed, many oncogenes code for either receptor or cellular tyrosine kinases. Because of its deregulation in many cancers (bladder, breast, cervix, colon, esophagus, head and neck, lung, and prostate), the epidermal growth factor receptor (EGFR) has been selected as a potential target for chemoprevention. Because growth factor networks are redundant, selective inhibition of signaling pathways activated in precancerous and cancerous cells should be possible. Requirements for specific EGFR inhibitors include specificity for EGFR, high potency, activity in intact cells, and activity in vivo. Inhibition of autophosphorylation is preferred, because it should result in total blockade of the signaling pathway. Inhibitors that compete with substrate rather than at the ATP-binding site are also preferable, because they are not as likely to inhibit other ATP-using cellular enzymes. Several classes of specific EGFR inhibitors have been synthesized recently, including structures such as benzylidene malononitriles, dianilinophthalimides, quinazolines, pyrimidines, [(alkylamino)methyl]-acrylophenones, enollactones, dihydroxybenzylaminosalicylates, 2-thioindoles, aminoflavones, and tyrosine analogue-containing peptides. A possible testing strategy for the development of these and other EGFR inhibitors as chemopreventive agents includes the following steps: (a) determine EGFR tyrosine kinase inhibitory activity in vitro; (b) evaluate EGFR specificity and selectivity (relative to other tyrosine kinases and other protein kinases); (c) determine inhibition of EGFR-mediated effects in intact cells; (d) determine inhibition of EGFR-mediated effects in vivo (e.g., in nude mouse tumor xenografts); and (e) determine chemopreventive efficacy in vivo (e.g., in the hamster buccal pouch or mouse or rat bladder).

Aromatase inhibitors as potential cancer chemopreventives.

Epidemiological and experimental evidence strongly supports a role for estrogens in the development and growth of breast tumors. A role for estrogen in prostate neoplasia has also been postulated. Therefore, one chemopreventive strategy for breast and prostate cancers is to decrease estrogen production. This can be accomplished by inhibiting aromatase, the enzyme that catalyzes the final, rate-limiting step in estrogen biosynthesis. The use of aromatase inhibitors is of clinical interest for cancer therapy, and selective, potent aromatase inhibitors have been developed. Several of these agents have demonstrated chemopreventive efficacy in animal models. The rationale for the use of aromatase inhibitors as chemopreventives and identification of inhibitors to serve as potential chemopreventive agents are the subjects of this review. After background information regarding aromatase is presented, the data for each inhibitor are summarized separately. The discussion focuses on those inhibitors that are clinically available or in clinical trials, including: aminoglutethimide (Cytadren), rogletimide, fadrozole hydrochloride, liarozole hydrochloride, anastrozole (Arimidex), letrozole, vorozole, formestane, exemestane, and atamestane. On the basis of results from preclinical studies, aromatase inhibitors may be promising agents for clinical trials in populations at high risk for developing estrogen-dependent cancers. Total suppression of aromatase may have adverse effects, as is evident in postmenopausal women (increased osteoporosis, cardiovascular disease, and urogenital atrophy). However, on the basis of preclinical studies of chemopreventive efficacy and chemotherapeutic applications of aromatase inhibitors showing dose-response efficacy, it may be possible to obtain chemopreventive effects without total suppression of aromatase and circulating estrogen levels. Suppressing local estrogen production may be an alternative strategy, as suggested by the discovery of a unique transcriptional promoter of aromatase gene expression, I.4, in breast adipose tissue. The development of drugs that target this promoter region may be possible.

Farnesyl protein transferase inhibitors as potential cancer chemopreventives.

Among the most important targets for chemopreventive intervention and drug development are deregulated signal transduction pathways. Ras proteins serve as central connectors between signals generated at the plasma membrane and nuclear effectors; thus, disrupting the Ras signaling pathway could have significant potential as a cancer chemopreventive strategy. Target organs for Ras-based chemopreventive strategies include those associated with activating ras mutations (e.g., colorectum, pancreas, and lung) and those carrying aberrations in upstream element(s), such as growth factors and their receptors. Ras proteins require posttranslational modification with a farnesyl moiety for both normal and oncogenic activity. Inhibitors of the enzyme that catalyzes this reaction, farnesyl protein transferase (FPT) should, therefore, inhibit Ras-dependent proliferative activity in cancerous and precancerous lesions (J. B. Gibbs et al., Cell, 77: 175-178, 1994). Because growth factor networks are redundant, selective inhibition of signaling pathways activated in precancerous and cancerous cells should be possible. Requirements for Ras farnesylation inhibitors include: specificity for FPT compared with other prenyl transferases; specificity for FPT compared with other farnesyl PPi-utilizing enzymes; ability to specifically inhibit processing of mutant K-ras (the most commonly mutated ras gene in human cancers); high potency; selective activity in intact cells; activity in vivo; and lack of toxicity. Numerous FPT inhibitors have been identified through random screening of natural products and by rational design of analogues of the two substrates, farnesyl PPi and the COOH-terminal CAAX motif of Ras tetrapeptides. A possible testing strategy for developing FPT inhibitors as chemopreventive agents includes the following steps: (a) determine FPT inhibitory activity in vitro; (b) evaluate selectivity (relative to other protein prenyl transferases and FPT-utilizing enzymes); (c) determine inhibition of Ras-mediated effects in intact cells; (d) determine inhibition of Ras-mediated effects in vivo (e.g., in nude mouse tumor xenografts); and (e) determine chemopreventive efficacy in vivo (e.g., in carcinogen-induced A/J mouse lung, rat colon, or hamster pancreas).

Lipoxygenase inhibitors as potential cancer chemopreventives.

Mounting evidence suggests that lipoxygenase (LO)-catalyzed products have a profound influence on the development and progression of human cancers. Compared with normal tissues, significantly elevated levels of LO metabolites have been found in lung, prostate, breast, colon, and skin cancer cells, as well as in cells from patients with both acute and chronic leukemias. LO-mediated products elicit diverse biological activities needed for neoplastic cell growth, influencing growth factor and transcription factor activation, oncogene induction, stimulation of tumor cell adhesion, and regulation of apoptotic cell death. Agents that block LO-catalyzed activity may be effective in preventing cancer by interfering with signaling events needed for tumor growth. In fact, in a few studies, LO inhibitors have prevented carcinogen-induced lung adenomas and rat mammary gland cancers. During the past 10 years, pharmacological agents that specifically inhibit the LO-mediated signaling pathways are now commercially available to treat inflammatory diseases such as asthma, arthritis, and psoriasis. These well-characterized agents, representing two general drug effect mechanisms, are considered good candidates for clinical chemoprevention studies. One mechanism is inhibition of LO activity (5-LO and associated enzymes, or 12-LO); the second is leukotriene receptor antagonism. Although the receptor antagonists have high potential in treating asthma and other diseases where drug effects are clearly mediated by the leukotriene receptors, enzyme activity inhibitors may be better candidates for chemopreventive intervention, because inhibition of these enzymes directly reduces fatty acid metabolite production, with concomitant damping of the associated inflammatory, proliferative, and metastatic activities that contribute to carcinogenesis. However, because receptor antagonists have aerosol formulations and possible antiproliferative activity, they may also have potential, particularly in the lung, where topical application of such formulations is feasible.

Perspectives on surrogate end points in the development of drugs that reduce the risk of cancer.

This paper proposes a scientific basis and possible strategy for applying surrogate end points in chemopreventive drug development. The potential surrogate end points for cancer incidence described are both phenotypic (at the tissue, cellular, and molecular levels) and genotypic biomarkers. To establish chemopreventive efficacy in randomized, placebo-controlled clinical trials, it is expected that in most cases it will be critical to ensure that virtually all of the biomarker lesions are prevented or that the lesions prevented are those with the potential to progress. This would require that both the phenotype and genotype of the target tissue in agent-treated subjects, especially in any new or remaining precancers, are equivalent to or show less progression than those of placebo-treated subjects. In the National Cancer Institute chemoprevention program, histological modulation of a precancer (intraepithelial neoplasia) has thus far been the primary phenotypic surrogate end point in chemoprevention trials. Additionally, we give high priority to biomarkers measuring specific and general genotypic changes correlating to the carcinogenesis progression model for the targeted cancer (e.g., progressive genomic instability as measured by loss of heterozygosity or amplification at a specific microsatellite loci). Other potential surrogate end points that may occur earlier in carcinogenesis are being analyzed in these precancers and in nearby normal appearing tissues. These biomarkers include proliferation and differentiation indices, specific gene and general chromosome damage, cell growth regulatory molecules, and biochemical activities (e.g., enzyme inhibition). Serum biomarkers also may be monitored (e.g., prostate-specific antigen) because of their accessibility. Potentially chemopreventive drug effects of the test agent also may be measured (e.g., tissue and serum estrogen levels in studies of steroid aromatase inhibitors). These initial studies are expected to expand the list of validated surrogate end points for future use. Continued discussion and research among the National Cancer Institute, the Food and Drug Administration, industry, and academia are needed to ensure that surrogate end point-based chemoprevention indications are feasible.

Cancer chemoprevention: progress and promise.

Cancer chemoprevention is the use of agents to inhibit, delay or reverse carcinogenesis. The focus of chemoprevention research in the next millennium will include defining the genotypic and phenotypic (functional and histological) changes during carcinogenesis, the cancer risk conferred by these changes, their modulation in preclinical experimentation and randomised clinical trials by chemopreventive drugs, dietary agents and regimens and treatments resulting from early detection. The key elements of this research effort will be basic and translational risk evaluation programmes; chemopreventive and dietary agent drug discovery and development; development of transgenic animal models; required safety and pharmacology studies; well-designed phase I, II and III chemoprevention studies; and much expanded early detection programmes. The large number of chemoprevention research programmes now ongoing ensures that the promise of chemoprevention will continue to be realised in the next decade.

Cancer chemoprevention: progress and promise.

Cancer chemoprevention is the use of agents to inhibit, delay or reverse carcinogenesis. The focus of chemoprevention research in the next millennium will include defining the genotypic and phenotypic (functional and histological) changes during carcinogenesis, the cancer risk conferred by these changes, their modulation in preclinical experimentation and randomised clinical trials by chemopreventive drugs, dietary agents and regimens and treatments resulting from early detection. The key elements of this research effort will be basic and translational risk evaluation programmes; chemopreventive and dietary agent drug discovery and development; development of transgenic animal models; required safety and pharmacology studies; well-designed phase I, II and III chemoprevention studies; and much expanded early detection programmes. The large number of chemoprevention research programmes now ongoing ensures that the promise of chemoprevention will continue to be realised in the next decade.

Progress in clinical chemoprevention.

Chemoprevention has four goals: (1) inhibition of carcinogens, (2) logical intervention for persons at genetic risk for cancer, (3) treatment of precancerous lesions, and (4) confirmation and translation of leads from dietary epidemiology into intervention strategies. The National Cancer Institute has described a multidisciplinary, cancer science-based program for chemopreventive drug development that addresses these objectives, and has collaborated with the US Food and Drug Administration to provide consensus guidance for applying this approach. A critical component is the identification and characterization of intermediate biomarkers of cancer and their validation as surrogate end points for cancer incidence in clinical chemoprevention trials. More than 40 agents in the program are currently on the clinical development path (preclinical toxicology and phase I clinical safety studies or phase II/III efficacy trials), with the major effort in phase II studies to identify and characterize intermediate biomarkers. The continually advancing knowledge of molecular and tissue-based carcinogenesis mechanisms will provide leads to new chemopreventive agents with increased specificity for carcinogenesis-related activities and, hence, reduced toxicity by virtue of minimal effects on normal cell and tissue functions. Results from the Human Genome Project will help identify and evaluate the potential for chemopreventive intervention in cohorts at genetic risk and will provide specific target lesions for intervention strategies.

Potential use of lipoxygenase inhibitors for cancer chemoprevention.

Increasing evidence suggests that lipoxygenase (LO)-catalysed metabolites have a profound influence on the development and progression of human cancers. Compared with normal tissues, significantly elevated levels of LO products have been found in breast tumours, colon cancers, lung, skin and prostate cancers, as well as in cells from patients with both acute and chronic leukaemias. LO-mediated products elicit diverse biological activities needed for neoplastic cell growth, influencing growth factor and transcription factor activation, oncogene induction, stimulation of tumour cell adhesion and regulation of apoptotic cell death. Agents that block LO catalytic activity may be effective in preventing cancer by interfering with signalling events needed for tumour growth. In the past ten years, pharmaceuticals agents that specifically inhibit the 5-LO metabolic pathway have been developed to treat inflammatory diseases such as asthma, arthritis and psoriasis. Some of these compounds possess anti-oxidant properties and may be effective in preventing cancer by blocking free radical-induced genetic damage or by preventing the metabolic activation of carcinogens. Other compounds may work by negatively modulating DNA synthesis. Pharmacological profiles of potential chemopreventive agents are compiled from enzyme assays, in vitro testing (e.g., cell proliferation inhibition in human cancer cells) and in vivo animal carcinogenesis models (e.g., N-methyl-N-nitrosourea-induced rat mammary cancer, benzo(a)pyrene-induced lung tumours in strain A/J mice and hormone-induced prostate tumours in rats). In this way, compounds are identified for chemoprevention trials in human subjects. Based on currently available data, it is expected that the prevention of lung and prostate cancer will be initially studied in human trials of LO inhibitors.

Progress in cancer chemoprevention.

More than 40 promising agents and agent combinations are being evaluated clinically as chemopreventive drugs for major cancer targets. A few have been in vanguard, large-scale intervention trials--for example, the studies of tamoxifen and fenretinide in breast, 13-cis-retinoic acid in head and neck, vitamin E and selenium in prostate, and calcium in colon. These and other agents are currently in phase II chemoprevention trials to establish the scope of their chemopreventive efficacy and to develop intermediate biomarkers as surrogate end points for cancer incidence in future studies. In this group are fenretinide, 2-difluoromethylornithine, and oltipraz. Nonsteroidal anti-inflammatories (NSAID) are also in this group because of their colon cancer chemopreventive effects in clinical intervention, epidemiological, and animal studies. New agents are continually considered for development as chemopreventive drugs. Preventive strategies with antiandrogens are evolving for prostate cancer. Anti-inflammatories that selectively inhibit inducible cyclooxygenase (COX)-2 are being investigated in colon as alternatives to the NSAID, which inhibit both COX-1 and COX-2 and derive their toxicity from COX-1 inhibition. Newer retinoids with reduced toxicity, increased efficacy, or both (e.g., 9-cis-retinoic acid) are being investigated. Promising chemopreventive drugs are also being developed from dietary substances (e.g., green and black tea polyphenols, soy isoflavones, curcumin, phenethyl isothiocyanate, sulforaphane, lycopene, indole-3-carbinol, perillyl alcohol). Basic and translational research necessary to progress in chemopreventive agent development includes, for example, (1) molecular and genomic biomarkers that can be used for risk assessment and as surrogate end points in clinical studies, (2) animal carcinogenesis models that mimic human disease (including transgenic and gene knockout mice), and (3) novel agent treatment regimens (e.g., local delivery to cancer targets, agent combinations, and pharmacodynamically guided dosing).

Preclinical drug development paradigms for chemopreventives.

Preclinical screening studies and animal efficacy testing models currently are used by the National Cancer Institute's chemoprevention drug discovery program to assess and identify chemical agents and natural products that may have the potential to prevent human cancer. Identification of potential cancer preventing agents begins by subjecting each compound to a sequential series of short-term, in vitro prescreens of mechanistic, biochemical assays to provide quantitative data to help establish an early indication of chemopreventive efficacy and to assist in prioritizing agents for further evaluation in longer-term, in vitro transformation bioassays and whole animal models. Promising chemical agents or combinations of agents that work through different inhibitory mechanisms subsequently are tested in well-established, chemically induced, animal tumor models, which include models of the lung, bladder, mammaries, prostate, and skin. These preclinical bioassays afford a strategic framework for evaluating agents according to defined criteria, and not only provide evidence of agent efficacy, but also serve to generate valuable dose-response, toxicity, and pharmacokinetic data required prior to phase I clinical safety testing. Based on preclinical efficacy and toxicity screening studies, only the most successful agents considered to have potential as human chemopreventives progress into clinical chemoprevention trials.

Chemoprevention of OH-BBN-induced bladder cancer in mice by oltipraz, alone and in combination with 4-HPR and DFMO.

The chemopreventive efficacy of the schistosomicidal drug oltipraz (5-(2-pyrazinyl)-4-methyl-1,2-dithiol-3-thione) was evaluated against urinary bladder transitional cell carcinoma (TCC) induced in male C57BL/6 x DBA/2FI (BDF) mice by N-butyl-N(4-hydroxybutyl)nitrosamine (OH-BBN). Oltipraz was fed in the diet from one week prior to OH-BBN dosing until sacrifice, six months later. The agent at 250 mg/kg diet significantly reduced the incidence of TCC compared with that in carcinogen controls. Oltipraz also significantly reduced TCC incidence when fed at 500 mg/kg diet for 76 days, then at 125 mg/kg diet until the end of the test period. Treatment with this higher dose level of oltipraz also appeared to decreases the depth of tumor invasion. At lower dose levels of 100 and 200 mg/kg diet, oltipraz alone had no effect on tumor incidence. It also was tested at these dose levels in combinations with 2-difluoromethylornithine (DFMO) and with all-trans-N-(4-hydroxyphenyl)retinamide (4-HPR). Treatment with the combination of 640 mg DFMO/kg and 100 mg oltipraz/kg diet was efficacious, although DFMO alone at 640 mg/kg diet was inactive. The combination of 1280 mg DFMO/kg and 200 mg oltipraz/kg diet reduced TCC incidence significantly compared with carcinogen controls, but the effect was no greater than that of DFMO alone at 1280 mg/kg, and weight gain was suppressed compared with carcinogen controls. The depth of tumor invasion was decreased with this combination treatment. Combinations of oltipraz at 100 and 200 mg/kg diet, 4-HPR at 156 and 313 mg/kg diet, and DFMO at 640 and 1280 mg/kg diet were efficacious and without apparent toxicity. Nonetheless, the three agent combinations cannot be considered more effective than DFMO alone at 1280 mg/kg diet or the lower dose combination of oltipraz and DFMO.

Chemoprevention of MNU-induced mammary tumors in the mature rat by 4-HPR and tamoxifen.

The chemopreventive efficacies of the retinoid all-trans-N-(4-hydroxyphenyl)-retinamide (4-HPR) and the anti-estrogen tamoxifen citrate were evaluated against N-methyl-N'-nitrosourea (MNU) induced mammary cancer in 120-day old female Sprague-Dawley rats. The agents were tested alone and in combination. They were administered in a modified AIN-76A diet, beginning 60 days prior to a single i.v. dose of 50 mg MNU/kg-bw and continuing until the end of the study, 180 days post-carcinogen treatment. At 782 mg/kg diet, 4-HPR alone significantly inhibited the induction of mammary adenocarcinomas compared with carcinogen controls. At 0.250 mg/kg diet, tamoxifen alone reduced tumor incidence compared with carcinogen controls. At 0.125 mg/kg diet, tamoxifen was ineffective. Combinations of 782 mg 4-HPR/kg diet with either 0.250 or 0.125 mg tamoxifen/kg diet were effective in inhibiting MNU-induced adenocarcinomas. The reductions in tumor incidence were greater for these combinations than for either agent alone. 4-HPR and 0.250 mg tamoxifen/kg diet decreased tumor incidence 81% (p less than 0.005), whereas 4-HPR and 0.125 mg tamoxifen/kg diet decreased tumor incidence 72% (p less than 0.005) compared with carcinogen controls. The combination of 391 mg 4-HPR/kg diet and 0.500 mg tamoxifen/kg diet was also tested and was effective in reducing tumor incidence.

A quantitative structure-toxicity relationships model for the dermal sensitization guinea pig maximization assay.

We have developed quantitative structure-toxicity relationship (QSTR) models for assessing dermal sensitization using guinea pig maximization test (GPMT) results. The models are derived from 315 carefully evaluated chemicals. There are two models, one for aromatics (excluding one-benzene-ring compounds), and the other for aliphatics and one-benzene-ring compounds. For sensitizers, the models can resolve whether they are weak/moderate or severe sensitizers. The statistical methodology, based on linear discriminant analysis, incorporates an optimum prediction space (OPS) algorithm. This algorithm ensures that the QSTR model will be used only to make predictions on query structures which fall within its domain. Calculation of the similarities between a query structure and the database compounds from which the applicable model was developed are used to validate each skin sensitization assessment. The cross-validated specificity of the equations ranges between 81 and 91%, and the sensitivity between 85 and 95%. For an independent test set, specificity is 79%, and sensitivity 82%.