Peptide-based Higher Order Structure (HOS) Sameness

In 2021, the Food and Drug Administration (FDA) issued its ANDA Guidance for Certain Highly Purified Synthetic Peptide Drug Products that Refer to Listed Drugs of rDNA Origin. This document outlines the requirements for ANDA submissions regarding complex active pharmaceutical ingredient (API) sameness relative to established products. In addition, recent journal publications by scientists associated with the FDA contain best practice recommendations to be used when evaluating generic and reference listed drug (RLD) product sameness in ANDA applications of peptide-based therapeutics. These publications, commonly referred to by reviewing chemists, can be summarized as: 

• Mass spectrometry (MS) to determine the peptide’s primary sequence and impurity profile (See Protein MS for additional details of Novatia’s Protein / Peptide MS services)
• One-dimensional (1D) and two-dimensional (2D) nuclear magnetic resonance (NMR) spectroscopy to characterize HOS (described below)
• Size-exclusion chromatography (SEC), dynamic light scattering (DLS) and diffusion ordered spectroscopy (DOSY) NMR to characterize aggregation (see Aggregation by DLS and Characterization of mAbs by NMR and DLS for additional details of Novatia’s Aggregation services).
•  

For peptides and proteins, the chemical shifts of backbone amide nitrogen and proton resonances are very sensitive to primary, secondary, and tertiary structure. Small changes in the local environment can affect these backbone chemical shifts. These structural differences have been well established in larger proteins, such as ribonuclease A, where the nitrogen – proton resonance pairs for the native protein appear visually different from the resonances of the SS scrambled disulfides (See our Biologics and Biosimilars page for more details on the structural differences in native and SS scrambled Ribonuclease A). With the emergence of more and more peptide-based therapeutics, the same concepts apply, for example, spectra of two common peptide-based therapeutics are shown in Figure 1 below. The pattern of chemical shifts in the amide (N-H) region of a 1D ¹H spectrum (6-11 ppm) is considered as a fingerprint of each peptide. Thus, the 1D ¹H NMR can qualitatively differentiate between two discrete peptide-based therapeutics (as shown in Figure 1) or quantitatively establish API HOS sameness of a generic peptide-based therapeutic to an RLD.

Figure 1: Stacked spectra illustrating the difference in 1D ¹H NMR spectra of two common peptide-based therapeutics. *Formulation excipients cut from spectrum for clarity.

The use of principal components analysis (PCA) on the 1D ¹H NMR spectra has been the recommended method for quantitatively establishing HOS sameness. The resulting Mahalanobis distance of the first three principal components (PCs) combined must not exceed 3.3 to establish sameness. An example of PCA where HOS sameness is and is not established is shown in Figure 2 below. However, different peptides are not the only factor that can affect PCA. The Novatia team has recently published in AAPS about the effects of pH on PCA (See Effect of pH on Peptide HOS Sameness for additional details).

Figure 2: A.) Overlaid spectra illustrating the difference in 1D ¹H NMR spectra of two different peptide-based therapeutics. Inset: PC Scores Plot for PC1 and PC2 which account for the two PCs that explain the most variance.  D_M (PCs 1 – 3) = 204.7, which exceeds the D_M < 3.3 threshold for Sameness. B.) Overlaid spectra illustrating the difference in 1D ¹H NMR spectra of two of same peptide-based therapeutics from different manufacturers. Inset: PC Scores Plot for PC1 and PC2 which account for the two PCs that explain the most variance. D_M (PCs 1 – 3) = 1.01, which meets the D_M < 3.3 threshold for Sameness. *Formulation excipients cut from spectrum for clarity.

In addition to 1D NMR PCA comparisons, the use of 2D NMR can provide additional statistical figures of merit when evaluating HOS Sameness by comparing the absolute peak height and combined chemical shift deviation (CCSD) of each peak in the generic and RLD products. The 2D data from the same peptide-based product from two different manufacturers are considered the same when:

• A t-test on the absolute peak heights of each peak with a signal-to-noise ratio (SNR) > 5 in all samples results in a p-value > 0.05, and,
• The overall average CCSD of all peaks with a SNR > 5 is less than 15 ppb for ¹H-¹³C 2D data or 8 ppb for ¹H-¹⁵N 2D data.

An example of the peak intensity and CCSD comparison is shown in Figure 3.

Figure 3: A.) Overlaid 2D ¹H-¹⁵N HMQC SOFAST Spectra of two different brands of the same peptide-based therapeutic. B.) Absolute Peak Height Comparison for all peaks with a SNR > 5 of two brands of the same peptide-based therapeutic. Yellow box indicates the one peak with a p < 0.05 for intensity similarity. C.) Individual peak CCSD comparison for all peaks with a SNR > 5. D.) Average CCSD for all peaks with SNR > 5. Red line indicates the < 8 ppb threshold for sameness.