Characterization of antisense oligonucleotides by Ion Pair HPLC – High Resolution MS: method development using Design of Experiments

Since their introduction in 1998 with the approval of Fomiversen for the treatment of CMV retinitis, Antisense Oligonucleotides (ASO) have been finding growing use in the therapy of genetic, oncologic, inflammatory and viral diseases. Chemically, they are short sequences of single-stranded RNA/DNA nucleotides, prepared by solid-phase synthesis (SPS) and modified to improve stability, pharmacokinetic properties and binding affinity; they have base sequences complementary to a defined mRNA sequence, to which they bind with high specificity inhibiting the expression of genes involved in the targeted disease. Their molecular mass normally ranges from 6 to 18 kDa, somewhat midway between small molecules and most protein therapeutics; given the structural similarity to their biological counterparts, and at the same time their synthetic origin, from a Regulatory point of view ASO cannot be considered as biotherapeutics, nor as small-molecule drugs. Consequently, application of the existing guidelines, i.e. those defining manufacturing process, characterization, specifications and analytical control is not always possible or appropriate. In the effort to fill the gap, several position papers and, very recently, a draft guideline from European Medicines Agency have been issued. As for any new drug under development, identification and quantification of the product-related impurities are key analytical issues, strictly connected not only to quality, but even more to safety for patients (the occurring impurities do not have to alter the drug pharmaco-toxicological profile) and efficacy (the total amount of impurities determines the assay potency of the drug substance). Indeed, the characterization and analysis of ASO drug substances, drug products and their impurities pose specific challenges:  the preparation by SPS involves tens of iterative steps, generally without purification; this produces a very complex impurity profile; consequently, the impurities are usually considered collectively in different classes (n-1, n+1, phosphate, abasic etc.), rather than individual impurities as with small drug molecules;  impurities are closely related to each other and to the full-length oligonucleotide: they have very similar physico-chemical properties, molecular mass and same UV absorption maxima (260 nm);  due to the presence of an acidic phosphorotioate (or phosphate) group for each residue, ASO have a net negative charge and are very polar: hence, they are difficult to separate chromatographically. Altogether, difficulty in separation and complex impurity profiles render unavoidable that many impurities are unseparated from the main product and from each other, requiring the use of orthogonal separation techniques (e.g. Anion Exchange HPLC and Ion Pair Reverse Phase HPLC) and/or orthogonal detection methods (e.g. HPLC-UV/MS), with complex method setup and data processing procedures. In the effort of studying a general and possibly simplified procedure for the analysis of therapeutic oligonucleotides and impurities, we developed an Ion Pair Reverse Phase HPLC – High Resolution MS method using Fomiversen (cited above) and Tofersen (used for the therapy of SLA) as model compounds, with a Thermo Vanquish UHPLC chromatograph interfaced to a Thermo Q Exactive Plus mass spectrometer. In the preliminary experiments, an OFAT (One Factor At a Time) approach was used to optimize mass parameters and determine which ion-pairing agent was the most appropriate for chromatography. In this first phase, the research focused on the intensity of the height of the MS signal and the chromatographic peak shape, and the following optimized values, valid for both model compounds, were selected: type of ion-coupling agent, N,N-diisopropylethylamine; Sheath Gas, 60 units; Auxiliary Gas Temperature, 400 °C; S-lens, 90 units. Subsequently, optimization continued through Design of Experiments (DoE), utilizing Response Surface Methodology (RSM) with Box-Behnken matrix. The considered factors in the RSM included the concentrations of N,N-diisopropylethylamine and hexafluoroisopropanol in the mobile phase, as well as the elution gradient slope. The evaluated responses included height and width of the main peak (from the UV trace), as well as critical resolutions between selected impurities (from their extracted MS chromatograms). The hypothesized models were found to be valid and significant with good prediction quality, and through RSM it was possible to define a multidimensional space (sweet spot) where the requirements for the responses considered were met, performing their simultaneous optimization. In conclusion, two distinct IP RP HPLC-MS methods were developed for Fomiversen and Tofersen, based on common mobile phase components and MS parameters, and on DoE application for the chromatographic parameters and UV signal optimization. In addition, 9 out of 11 detected Tofersen impurities above 0.1% were identified, whereas 3 of 4 Fomiversen impurities were identified.