Inter-laboratory comparison of water solubility methods applied to difficult-to-test substances

Abstract

Water solubility is perhaps the single most important physical–chemical property determining the environmental fate and effects of organic compounds. Its determination is particularly challenging for compounds with extremely low solubility, frequently referred to as “difficult-to-test” substances and having solubility’s generally less than 0.1 mg/L. The existing regulatory water solubility test for these compounds is the column elution method. Its applicability, however, is limited, to non-volatile solid or crystalline hydrophobic organic compounds. There currently exists no test guideline for measuring the water solubility of very hydrophobic liquid, and potentially volatile, difficult-to-test compounds. This paper describes a “slow-stir” water solubility methodology along with results of a ring trial across five laboratories evaluating the method’s performance. The slow-stir method was applied to n-hexylcyclohexane, a volatile, liquid hydrophobic hydrocarbon. In order to benchmark the inter-laboratory variability associated with the proposed slow-stir method, the five laboratories separately determined the solubility of dodecahydrotriphenylene, a hydrophobic solid compound using the existing column elution guideline. Results across the participating laboratories indicated comparable reproducibility with relative standard deviations (RSD) of 20% or less reported for each test compound – solubility method pair. The inter-laboratory RSD was 16% for n-hexylcyclohexane (mean 14 µg/L, n = 5) using the slow-stir method. For dodecahydrotriphenylene, the inter-laboratory RSD was 20% (mean 2.6 µg/L, n = 4) using the existing column elution method. This study outlines approaches that should be followed and the experimental parameters that have been deemed important for an expanded ring trial of the slow-stir water solubility method.

Evaluation of Deep Learning Architectures for Aqueous Solubility Prediction

“…However, we should also keep in mind that the predictive task is performed on log solubility, which means that an absolute error of 2 orders of magnitude represents a much smaller actual error for low solubility bins than it does for high solubility bins. It is also known that experimental limitations make it challenging to measure very low solubility values 71 and these measurements are likely associated with high uncertainties.…”

Evaluation of Machine Learning Models for Aqueous Solubility Prediction in Drug Discovery

“…While experimental solubility measurements are resource-intensive and time-consuming, in-silico prediction methods offer a promising alternative [4]. By harnessing the wealth of information encoded within molecular structures, these methods offer a powerful approach to estimating a compound’s aqueous solubility without the need for laboratory experiments [5]. However, the accurate and rapid prediction of aqueous solubility has posed a significant challenge in drug discovery for decades.…”

Transparency in Modeling through Careful Application of OECD’s QSAR/QSPR Principles via a Curated Water Solubility Data Set

“…For pharmaceutical development, accurate water solubilities are often needed to optimize candidate drugs to possess suitable transport properties within tissues of interest, with hydrophilic (high water solubility) species possessing differing localization and metabolic properties than their hydrophobic (low water solubility) counterparts. Reliable water solubility values play an important role in reducing the costs and enquiry space surrounding expensive and potentially challenging (i.e., difficult-to-test substances) empirical investigations …”

“…Reliable water solubility values play an important role in reducing the costs and enquiry space surrounding expensive and potentially challenging (i.e., difficult-to-test substances) empirical investigations. 4 T h i s c o n t e n t i s As a thermodynamic property, the value of water solubility depends on environmental conditions such as pressure and temperature. 5 It additionally depends on structural characteristics such as exposed van der Waals surface area, quantity of hydrogen-bond acceptors and donors, and acidity.…”