Using ARROWSMITH: a computer-assisted approach to formulating and assessing scientific hypotheses.

Conventional computer searches of the biomedical literature (e.g. MEDLINE) allow investigators to retrieve much of the information that has already been published on a given topic. However, these searches are of limited utility at the frontier of scientific discovery, when one wishes to identify and assess new, untested scientific hypotheses, or to uncover biologically significant relations between two previously disparate fields of inquiry. We have designed a set of interactive software and database search strategies, collectively called ARROWSMITH, that facilitate the discovery of plausible hypotheses linking findings across specialties (Artif. Intell. 91 (1997) 183-203). In the simplest implementation of ARROWSMITH, the user begins with an experimental finding or hypothesis that two items A and C are related in some way. The titles of papers indexed in MEDLINE which contain the word 'A' (or synonyms) are downloaded into a file A, and similarly a file C is created. The software constructs a list of words and phrases B common to files A and C; automatic and manual editing are used to filter out uninteresting B-terms. For each B-term, the software generates an AB file of titles containing both 'A' and 'B', and a BC file of titles containing both 'B' and 'C'; these titles are juxtaposed to facilitate the user judging whether there is likely to be a biologically significant relation among A, B and C. ARROWSMITH has been employed to analyze research problems relating to oxidative stress, brain damage, Alzheimer's disease and schizophrenia. Applications of ARROWSMITH include: anticipating adverse drug reactions, identifying mechanisms by which agents modulate cellular or organismal responses, suggesting new therapeutic approaches, identifying possible risk factors for diseases, and identifying potential animal models for human conditions. A simplified experimental version of ARROWSMITH is now freely accessible on the World Wide Web (http:@kiwi.uchicago.edu).

Linking estrogen to Alzheimer's disease: an informatics approach.

Epidemiologic studies suggest that estrogen protects against AD. We employ ARROWSMITH, a novel computer-assisted approach, to identify possible links between estrogen and AD that are not explicit in the biomedical literature, by searching for substances or processes that are known targets of estrogen action and that have also been separately studied in relation to AD. Several links appear particularly promising (e.g., estrogen's antioxidant activity) and merit attention by neuroscientists.

Medical literature as a potential source of new knowledge.

Specialized biomedical literatures have been found that are implicitly linked by arguments that they respectively contain, but which nonetheless do not cite or refer to each other. The combined arguments lead to new inferences and conclusions that cannot be drawn from the separate literatures. One such analysis identified one set of articles showing that dietary fish oils lead to certain blood and vascular changes, and a second set containing evidence that similar changes might benefit patients with Raynaud's syndrome. Yet these two literatures had no articles in common and had never before been cited together; neither literature mentioned the other or suggested that dietary fish oil might benefit Raynaud patients. Two years after publication of that analysis, the first clinical trial demonstrating such a beneficial effect was reported independently by others. A second example of literature synthesis, based on eleven indirect connections, led to an inference that magnesium deficiency might be a causal factor in migraine headache. A third example calls attention to implicit connections between arginine intake and blood levels of somatomedins, a potentially fruitful but neglected area of research with implications for the decline with age of thymic function and protein synthesis. A model and an online search strategy to aid in identifying other logically related noninteractive literatures is described. Such structures are probably not rare and may provide the foundation for a literature-based approach to scientific discovery.

Nifedipine gastrointestinal therapeutic system.

Convenient once-a-day dosage regimens are highly desirable in general, and especially for the treatment of asymptomatic diseases such as essential hypertension. Nifedipine is an insoluble, short-acting calcium channel blocker that presents a difficult technical challenge for formulation in a constant 24-hour delivery dosage form. Once-a-day dosage forms have been developed based on the gastrointestinal therapeutic system (GITS) push-pull osmotic pump configuration in three strengths with different drug delivery rates (mg/hour) per dose (mg), as 1.7/30, 3.4/60, and 5.1/90. The delivery rates of drug from these systems are controlled by their drug loading, composition of osmotic components, membrane properties, and dimensions. The release rates are independent of pH in the range from gastric pH = 1.2 to intestinal pH = 7.5. The release rates are independent of stirring rate and therefore unlikely to be influenced by motility in the gastrointestinal tract. The drug release rate from the nifedipine GITS dosage form in vivo in the gastrointestinal tract of dogs has been found to be equal to the release rate in vitro, indicating that the in vitro test is predictive of in vivo delivery. Following administration of the nifedipine GITS dosage forms to human subjects, absorption rates, calculated from resulting plasma concentrations, indicate that the cumulative amount of drug absorbed in humans over 24 hours is proportional to the amounts of drug delivered in vitro. Plasma concentrations are therefore predictable and remain relatively constant throughout the 24-hour dosing interval.

Osmotic delivery systems for the beta-adrenoceptor antagonists metoprolol and oxprenolol: design and evaluation of systems for once-daily administration.

The essential features and mode of action of oral osmotic drug delivery systems (Oros) for metoprolol fumarate and oxprenolol succinate are described. Critical aspects in the development of systems for once-daily administration of both drugs are discussed, and methods for evaluating in vitro release characteristics are presented. In vitro testing confirmed that drug delivery corresponded closely to the theoretical release behaviour predicted from the physicochemical and membrane permeability characteristics for both oxprenolol and metoprolol systems. In vitro release rates were also shown to be unaffected by pH, in vitro test procedures, dissolution media and long-term storage at different temperatures.

Evaluation of oxprenolol and metoprolol Oros systems in the dog: comparison of in vivo and in vitro drug release, and of drug absorption from duodenal and colonic infusion sites.

The performance of oxprenolol and metoprolol Oros systems has been evaluated in the dog. One study compared in vivo and in vitro release from both systems over 2-14 h. The other compared the systemic availabilities of both drugs after 3 h infusion at a constant rate into the cephalic and hepatic portal veins, and into the lumen of the duodenum and colon. In the in vivo release studies, Oros systems were recovered throughout the gut from the stomach to the colon. The amounts of drug remaining in the systems corresponded closely to those measured in a parallel in vitro release experiment. In vitro testing is thus a reliable indicator of in vivo system performance. In the absorption studies, both metoprolol and oxprenolol were shown to be subject to substantial first-pass metabolism. Additionally, for metoprolol the data indicated a significant loss during transport from the gut lumen into the portal circulation. For both drugs the availability from the colon was equal to that from the duodenum. These results provide some justification for the development of oral dosage forms with extended durations of release even for drugs which undergo significant first-pass metabolism.

Toxicity of anticholinesterases: interactions of pyridostigmine and physostigmine with soman.

This investigation was conducted to assess the potential of carbamate pretreatment to exacerbate the ill effects of low doses of soman. Ambulatory Activity in a photocell cage (AA) and performance time on an accelerating rotarod (ARR) were used to test for interactions between pyridostigmine or physostigmine and soman. ED50s (i.e., dosages sufficient to reduce ARR time and AA to 50% of control level) of each carbamate (IM) and soman (SC) were determined. The ED50 values (mg/kg) in the ARR test were 3.2, 0.21, and 0.072 for pyridostigmine, physostigmine and soman, respectively, while in the AA test the corresponding values were 1.8, 0.072 and 0.060. The matrix of 16 combinations of 0, 1, 2/3, and 1/3 ED50 each of carbamate and soman was studied in each test system, as well as the effect of behavioral deficit free (BDF) dosages of each carbamate on the ED50s of soman. In both the AA and ARR tests the matrix of combinations of pyridostigmine and soman indicated an additive effect. In contrast, physostigmine produced one instance of potentiation in each test system and anatagonism in two combinations in the AA procedure. A BDF dosage of each carbamate (0.056 mg/kg of pyridostigmine and 0.026 mg/kg of physostigmine) gave no evidence of adding to the deficit in AA induced by soman. In the ARR test, the ED50 of soman was lower by 11% with pyridostigmine pretreatment and by 14% with physostigmine; the latter just reached statistical significance (p less than 0.05). Although additivity was most often found at higher dosages of pyridostigmine and physostigmine, at the BDF dosages little or no adverse interaction was found between Pyridostigmine or Physostigmine and low levels of soman.