Huyi Zhang is chemist at the Division of Life-Cycle API, Office of New Drug Product, Office of Pharmaceutical Quality, Center of Drug Research and Evaluation Center, Food and Drug Administration.

Huyi Zhang

Type-II drug master files (DMFs) (1) are submissions of drug substance information to FDA to permit FDA to review this information in support of a third-party’s submission without revealing the information to the third party. A DMF must be adequate for referenced Investigational New Drug Application (IND), a New Drug Application (NDA), and an Abbreviated New Drug Application (ANDA) to be approved. The proof of drug substance identity in a Type II DMF is one of the key requirements by 21 

) 314.50(d)(1)(i) (2), and sections 3.2.S.3.1 and 3.2.S.5 of DMFs are dedicated to the 

Elucidation of Structure and Other Characteristics of the drug substance

. To establish the drug substance identity or in-house reference standards, the DMF holder needs to provide comprehensive structural elucidation information, including but not limited to infrared (IR) spectroscopy, nuclear magnetic resonance (NMR) spectroscopy, ultraviolet (UV) spectroscopy, mass spectrometry (MS), elemental analysis (EA), powder X-ray diffraction, differential scanning calorimetry (DSC), and specific optical rotation, where applicable. 
	
	The authors, in a series of articles, clarify the intent and criticality of the deficiencies cited in the DMF Sections 3.2.S.3.1 Elucidation of Structure and Other Characteristics and 3.2.S.5 Reference Standards or Materials. The authors have surveyed deficiency letters issued for 190 DMFs supporting ANDA submissions from January 2015 to September 2015. Most common deficiencies cited in sections 3.2.S.3.1 and 3.2.S.5 of the Common Technical Document (CTD) format are related to solution NMR data. Hence, this article focuses exclusively on the solution NMR portion of characterization data in the DMF submissions. Deficiencies related to other characterization methods will be discussed with detail in a separate article. This information will be useful for DMF holders to build submission quality into the DMFs by providing accurate and complete information, and subsequently reduce the DMF review time and accelerate the review process of ANDAs. 
	 

Deficiencies related to NMR, IR, UV, MS, and EA.

 Among the surveyed 190 DMFs, approximately 90% of the DMFs were for synthetic small-molecule drug substances and the remaining 10% were for semisynthetic antibiotics and peptides. Two hundred and seventy-one deficiencies related to NMR, IR, UV, MS, and EA were issued for Sections 3.2.S.3.1 and 3.2.S.5 in the DMF.  The distribution of deficiencies among NMR, IR, UV, MS, and EA methods and among the DMFs is provided in 

 presents the percentage of deficiencies for each method, calculated based on numbers of deficiencies for each method divided by the total numbers of deficiencies. The most prevalent NMR-related deficiencies followed by those related to IR, UV, EA, and MS are shown in a descending order. The respective percentile distributions per method were 55.7%, 17.0%, 12.5%, 10.0%, and 4.8%. 

 presents the percentage of DMFs with related deficiencies, calculated based on the numbers of DMFs with a specific category of deficiencies (one of these methods, NMR, IR, UV, EA, or MS) divided by total number of DMFs surveyed. The percentage numbers (76.8%, 23.7%, 17.9%, 14.2%, and 6.3%, respectively) reflect how often the DMFs have issues related to a given characterization method. Because a DMF may have deficiencies related to more than one method, it is reasonable to mention that the sum may exceed 100%.

 It is clear from the data that the deficiencies related to NMR are the most prevalent in DMF sections 3.2.S.3.1 and 3.2.S.5 with 55.7% NMR-related deficiencies and 76.8% of DMFs with at least one NMR-related deficiency. The NMR-related deficiencies are further classified into seven sub-categories. 

 presents the percentage of deficiency for each subcategory. 

 presents percentage of DMFs with related deficiencies in each subcategory. These two calculation approaches show the same trend of deficiency distribution. “No assignments for proton and/or carbon” is the most prevalent deficiency (23.2%/18.4%), followed by “Incorrect or incomplete assignments” (19.2%/14.7%), “No comparison to reference standards” (18.5%/14.2%), “Inadequate stereochemistry assignments” (17.9%/13.7%), “Miscellaneous issues (poor resolution, illegibility, incomplete information, etc.)” (9.3%/5.8%), “Not matched with reference standard” (7.3%/5.3%), and “Not well-resolved or overlapped spectra” (6.0%/4.7%). 

	The seven subcategory deficiencies related to NMR spectroscopy (see 

) are further discussed roughly in the descending order of percentage of DMFs. 
	
	

 This is the most commonly found deficiency related to NMR spectroscopy, and 18.4% of DMFs surveyed were observed to have such an issue. The DMFs included 

C NMR spectra of drug substances or in-house impurity reference standards, but did not provide any assignments on integration, position, or multiplicity of protons, and/or carbons, when applicable. 
	
	In the absence of adequate comparative data with a suitable reference standard, DMF holders should provide proton and carbon assignments to verify the proposed structure. Comparison with the literature data and/or authentic sample can be helpful and will be discussed later. The extent of spectrum interpretation should be contingent on the structural complexity and should adequately address issues such as constitutional isomers and stereoisomers. The DMF holders may choose appropriate techniques, including but not limited to 1D NMR (

C HMQC/HMBC, NOESY/ROESY, etc.) as fit for purposes. The DMF holders may choose to report the spectroscopic data following 

 (3). The solvent and instrument frequency should be identified. Unless greater precision is needed to distinguish closely spaced peaks, proton NMR shifts should be reported to 0.01 ppm precision and carbon NMR peak shifts should be reported to the nearest 0.1ppm. Tabulated assignments are recommended with chemical shifts including multiplicity, if applicable, 

 coupling constants if applicable and attribution as shown below in 

 Almost 15% of the DMFs surveyed provided partial or incorrect proton or carbon assignments for the drug substance or for the in-house impurity reference standards. This category does not include the deficiencies related to inadequate interpretation of stereoisomers, which will be discussed later. The most common deficiencies in this section include: assignment on multiplicity, integration, or position of proton not in agreement with the proposed structure, assignments incomplete for elucidating the proposed structure, or ignored or mistakenly assigned impurity peaks.  
	
	Incomplete or ambiguous spectroscopic analysis may lead to uncertain or even wrong structure assignments as exemplified with recently reported cases, Bosutinib (4) and ONC201 (5, 6, 7). For Bosutinib, a wrong regioisomer was picked for further exploration, while for ONC201, an incorrect skeletal isomer was assigned to enter IND process. Therefore, if applicable, it is recommended that DMF holders provide complete 

C  HMQC/HMBC interpretation to ensure the structure assignments. Comparison with literature interpretation, if applicable, is an alternative approach to confirm the structural assignments. Additionally, complete and accurate assignments of proton and carbons may help to identify organic impurities in the drug substance that may not be plausible by other analytical methods.
	
	The authors’ review experience show that high-performance liquid chromatography (HPLC) purity is not always consistent with NMR purity as certain organic impurities in drug substances may not be detected by UV because either the impurities have no UV absorption or the adopted UV wavelength is not able to detect the impurities. Further, because signal intensities of NMR spectroscopy are directly proportional to the number of nuclei either from drug substance or from impurities causing the signal (provided certain conditions are met), NMR spectroscopy can also serve a valuable orthogonal tool other than HPLC method to detect impurities.

 Approximately 14% of surveyed DMFs did not provide a comparison with available 

) reference standards, which are considered highly reliable for structural verification. Comparison of spectroscopic data of the proposed structures with those of high-purity reference standards is one of the most reliable and direct approaches to confirm the identity and purity of the drug substance. This approach is even more suitable for verifying complex structures associated with regioisomerism, geometric isomerism, and optical isomerism. Reference standards available from 

are considered highly reliable for structural verification. However, if the 

 reference standards are not available, the structure of other source reference standards should be unequivocally confirmed by a variety of spectroscopic methods to ensure the identity and purity. The authors recommend that the DMF holders provide spectroscopic data comparison as well as overlaid 

C NMR spectra of proposed structures and reference standards for better visual comparison.
	
	

 Spectra not matched with reference standards.

 More than 5% of surveyed DMF showed that the spectra of API or impurities are not matched exactly with those of reference standards in terms of chemical shifts, resolution, and/or splitting patterns (coupling constants). The most common deficiencies include: 

		Test samples and reference standards are not in the same form (i.e., free base and salt or two salts with different counter ions)

		NMR spectra of test sample and reference standard are taken in different deuterated solvents

		Usage of deuterated residual peak solvent instead of tetramethylsilane (TMS) as a reference

		Markedly different concentrations or different NMR running temperature of the test sample and corresponding reference standard.

	It is important to understand that both chemical shifts and coupling constants of a proton are variant properties of a molecule and may change depending on the molecular environment. Not only the chemical shifts of N-H and O-H but also the chemical shifts of C-H could be dependent on the nature of deuterated solvents (8, 9, 10), analyte concentration (11, 12, 13), 

 (14), and analyte state (free base vs. salt) (15) etc. Similarly, coupling constants are also solvent dependent (16).
	To achieve better matched NMR spectra, it is recommended that the following be considered: 

		Use of TMS instead of deuterated solvents as the reference standard

		Use of analytes in the same form (i.e., free base or the same salt)

		Use of a mixture of analyte and reference standard to completely eliminate the effects of solvent, concentration, temperature, etc. 

 Almost 14% of surveyed DMFs did not provide adequate assignment for stereoisomers. Considering not all the APIs have chiral centers, the actual percentage could be even higher if only chiral APIs are considered. Stereoisomers including geometric isomers, diastereoisomers, and enantiomers have the same molecular formula and sequence of bonded atoms, but differ in the dimensional orientations of their atoms in space (17).
	Regarding drug substances with stereoisomerism issues, the most common deficiencies include: 

		No convincing/sufficient proof to verify the relative configuration including but not limiting to, cis/trans geometry of double bond, equatorial/axial proton in a cyclohexane ring, cis/trans ring substituents, cis/trans fused rings, or endo/exo configuration

		No convincing/sufficient proof to establish absolute configuration of the chiral centers 

		No explicit data interpretation to address the stereochemistry issues.

	
	To address the above relative and absolute configuration issues, the relative configuration of API or impurities can be verified by comparison of NMR spectra of the in-house samples with those of the known reference standards. The overlaid spectra of the in-house samples and reference standards can be easily compared for better visualization. However, if reference standards are not available, NMR spectroscopy will be a powerful tool to determine relative configuration of molecules. Specifically, the vicinal coupling constants (

) (18), nuclear Overhauser effect (NOE) (19) and chemical shifts (20), can be used for determining relative configuration of olefins, geometric isomers, epimers of conformationally restrained scaffolds, or 1,3 diol derivatives (21, 22), etc.
	
	Usually the absolute configuration of a generic-drug substance or an impurity can be confirmed by comparison of the specific optical rotation with that of the known reference standard. However, for enantiomers with very small specific optical rotation, it might be inappropriate to verify the absolute configuration solely by specific optical rotation comparison. As an alternative, comparison of circular dichrosim (CD) or chiral HPLC retention time of the in-house samples with that of the known reference standard will give a higher level of confidence. In cases where reference standards are not available for comparison, NMR-based chiral derivatizing agent (CDA) method (23, 24, 25) can be a practical approach to determine absolute configurations as a complementary approach to X-ray. Regardless of the method, DMF holders should include legible copy of spectra with explicit interpretation. 
	
	

 Not well-resolved or overlapped spectra.

 In 4.7% of the surveyed DMFs, resonances of 

H NMR spectra are either overlapping or not well-resolved. The overlapping resonance lines or broad peaks makes it difficult to identify diagnostic chemical shifts (e.g., chemical shifts for NOE studies) and coupling constants (e.g., 

 for double bond geometry or relative configuration of cis-trans fused rings) to unequivocally assign the proposed structures. Poor solubility, concentrated samples, poor shimming, dynamics of conformationally flexible macrocycles (26), or protonated quaternary ammoniums (27) can be the common reasons for broad peaks without splitting details for small organic molecules.
	
	Such resonance broadening could be mitigated by better shimming, deuterated solvents with good sample solubility, diluting analyte concentration (11, 12, 13), heating the sample (variable temperature NMR) (14), or transforming a salt to a free base. A general strategy for resolving overlapping signals is to change the NMR solvent (8, 10, 28), change the temperature, or use chemical shift agents (29).
	
	

 Besides the previous specific categories, 5.8% of the DMF submissions provided incomplete or incorrect information. This incomplete information includes failure to provide batch numbers of samples/reference standards, submission of illegible spectra, and low observed signal/noise (S/N). 

	The authors have summarized the common deficiencies related to characterization of drug substances and related substances by solution NMR with reviewers’ perspective. The information provided herein is intended to assist DMF holders in providing sufficient characterization information with meaningful interpretation to establish the claimed identity, so that the number of deficiencies in this regard can be reduced. Common deficiencies and regulatory inputs may also apply for DMFs supporting INDs and NDAs due to the generalization. 

	This article represents the views of the authors and not of FDA.

	The authors thank Scott Furness, Ramesh Sood, Zhengfu Wang, Qian Shi, and Chandrawansha Senanayake for their valuable suggestions. They also acknowledge input from the DMF staff.
	 

, Title 21, Food and Drug (Government Printing Office, Washington, DC), Part 314, Volume 5, Subpart B, Chapter 1, Sec. 314.50.
	3. 

Notice to Authors of JACS Manuscripts, Compound Characterization and Computational Data

.
	4. B. Halford, “Bosutinib Buyer Beware,” 

, May 11, 2012.
	5. S. Borman, “Tug Of War Over Promising Cancer Drug Candidate,” 

, May 21, 2014.
	6. T. J. Nicholas, et al., 

53 (26), 6628-6631 (2014).
	7. Oncoceutics, “Oncoceutics Announces FDA Acceptance of Investigational New Drug Application for Phase I/II Trial of Novel Anti-Cancer Drug ONC201,” Press Release (Hummelstown, PA, Mar. 04, 2014. 
	8. H. Hoshina, et al., 

, 56 (19), 2941-2951 (2000).
	9. K. Kubo, et al., 

 37(33), 5917-5920 (1996).
	10. R. J. Abraham, et al., 

 44 (5), 491-509 (2006).
	11. A. Mitra, et al., 

, 54 (51), 15489-15498 (1998).
	12. J. A. Pople, W. G. Schneider, and H. J. Bernstein, 

High Resolution Nuclear Magnetic Resonance

 (McGraw-Hill Book Co., Inc., New York, N. Y., 1959), pp 424. 
	13. M.G. Reinecke, H.W. Johnson Jr., and J. F. Sebastian, 

 91 (14), 3817-3822 (1969).
	14. G. Facey, “

Variable Temperature to Improve NMR Resolution

. 51 (5), 292-8 (2013).
	16. M. Barfield and M. D. Johnston Jr. 

, 2nd ed. (the “Gold Book”) (1997). Online corrected version: (2006-).
	18. M. Karplus, 

 85 (18), 2870-2871 (1963).
	19. R. Kaiser, 

 39 (10), 2435-2442 (1963).
	20. A. G. Moritz and N. Sheppard, Mol. Phys. 5 (4), 361-368 (1962).
	21. S. D. Rychnovsky, B. Rogers, and G. Yang, 

., 58 (13), 3511-3515 (1993).
	22. D. A. Evans, D. L. Rieger and J. R. Gage, 

, 31 (49), 7099-7100 (1990).
	23. J. M. Seco, E. Quinoa, and R. Riguera, 

 104 (1), 17-118 (2004). 
	24. J. M. Seco, E. Quinoa, and R. Riguera, 

 112 (8), 4603-4641 (2012).
	25. T. J. Wenzel, and C.D. Cora, 

, 23 (3), 190-214 (2011).
	26. A. D. Bain, 

Progress in Nuclear Magnetic Resonance Spectroscopy

 43 (3-4), 63-103 (2003).
	27. R. Glaser, A. Peleg, and S. Geresh, 

.  28 (5), 389-396 (1990).
	28. J. Giner, et al., Magn. Reson. 

 45 (4), 351-354 (2007).
	29. J. K.M. Sanders, and D.H. Williams, 


	Vol. 41, No. 2
	February 2017
	Pages: 16-24

	When referring to this article, please cite it as J. Wang et al., "Characterization of Small-Molecule Drug Substances in Type II DMFs Supporting ANDAs-Part I: Solution Nuclear Magnetic Resonance Spectroscopy," 

Articles

The authors survey deficiency letters issued for 190 drug master files supporting abbreviated new drug application submissions.