Vial content uniformity and batch COA sampling: what a certificate does and does not say about the next vial in the box
A batch certificate of analysis carries a purity percentage, a stated fill mass, and — if the lab measured it — a fill accuracy. All three are batch-level statistics. None of them describe the specific vial in the researcher’s hand. Purity is measured on a pooled solution prepared from one or more sampled units. Fill accuracy is calculated from a small number of gravimetric determinations. The COA reports what the lab measured; the inference that the rest of the lot behaves the same way rests on how sampling was designed and on the pharmacopeial rules for judging uniformity across the batch.
For research work that depends on accurate mass per vial, the difference between batch mean and vial-to-vial variability is the difference between reproducible data and noise nobody attributes correctly. This article walks through what USP <905> actually requires, what the FDA’s lyophilization guidance adds, and what a research-grade COA does and does not certify. It closes on what our own certificates carry (our BPC-157 10 mg COA is used as the worked example) and — more usefully — what they do not.
The pharmacopeial framework: USP <905>
USP General Chapter <905> Uniformity of Dosage Units is the harmonized standard (also ICH Q4B Annex 6) for judging whether individual dosage units in a lot fall inside acceptable variability. It offers two methods. Weight Variation infers active content from unit weights combined with a batch assay; it is permitted where the active substance is at least 25 mg per unit and comprises at least 25% of the total unit mass. Content Uniformity measures actual drug content in each sampled unit by direct chemical analysis and is required for products that fall below those thresholds.
The staged acceptance rule is compact. Stage 1 tests ten units, computes an Acceptance Value AV = |M − X̄| + k × s (with k = 2.4), and passes if AV ≤ 15.0. Stage 2 tests a further twenty units — thirty total — recomputes AV with k = 2.0, and passes if the overall AV is still ≤ 15.0 and no single unit lies outside 0.75M to 1.25M of the reference. So the chapter defines an outlier boundary at ±25% and a spread boundary that scales with the standard deviation of the sample. It also defines what “sample” means: ten units at Stage 1, thirty at Stage 2, drawn to be representative of the lot.
That last word matters. A ten-unit AV computation is only as good as the sampling plan feeding it. Ten vials pulled consecutively from the same tray after fill do not sample vial-to-vial variability across a run in the way ten vials pulled at intervals across the whole batch do — a distinction the FDA’s 1993 inspection guide for lyophilized parenterals states directly.
Freeze-dried complications: weight is not content
The FDA inspection guide notes that weight variation “may be applied to solids…that have been prepared from true solutions and freeze-dried in final containers” but adds a caveat that catches manufacturers out: assaying vial contents without recording the sample weight “will provide information on the label claim of a product, but without knowing the sample weight will provide no information about dose uniformity.” Testing a vial with a higher fill volume yields a higher potency result and misrepresents batch potency in exactly the direction that hides fill variability.
For a lyophilized research peptide, three things vary vial-to-vial that a batch assay cannot separate:
- Fill volume before freezing. Peristaltic and time-pressure fillers have known coefficients of variation; a well-tuned line commonly runs at 1–2% CV, a poorly tuned line at 5% or higher. FDA MAPP 5019.1 requires that fill volumes not be “unreasonably large” (which would enable a second withdrawal) but permits enough excess to allow the labeled net content to be withdrawn — so an overfill deliberately built into the process must be quantified, not left implicit.
- Moisture after freeze-drying. Residual water in the cake affects both mass and long-term stability; a single-vial mass that appears normal can hide a wet vial with different real peptide content.
- Peptide identity across the fill. Solid-phase synthesis produces sequence-specific impurities that partition with the peptide through purification. If the pooled reference solution used for the assay is prepared from many vials, it averages those impurities out; the single vial being used in an experiment does not.
The consequence is arithmetic. A purity number certified against a pooled solution reports the composition of that pool. It is a reliable statement about the batch. It is a probabilistic statement about the vial.
Sampling design: the part a COA does not describe
A certificate reports what was measured. It rarely reports how many vials were sampled or how they were selected. That gap is where two lots with identical purity numbers can differ materially in real-world reproducibility.
The problem is not theoretical. In 2020 a joint U.S. Pharmacopeia collaborative study led by Li and colleagues at USP Biologics compared HPLC assay, quantitative NMR and amino acid analysis across multiple sites, finding that HPLC assay carried the lowest inter-lab variability while qNMR was flagged for further exploration — even for one homogeneous material, different methods give different absolute numbers (Li C et al., J Pharm Biomed Anal 166:105-112, 2019, PMID 30640042). The BIPM key comparison CCQM-K115.b on synthetic oxytocin ran the same experiment across eight national metrology institutes and returned individual results ranging from 766.3 mg/g to 987 mg/g on a single homogeneous batch — a spread of ~220 mg/g, or ~22 percentage points if expressed as purity. The single rejected result was the one that had not corrected for peptide-related impurities (assigned separately at 31.6 ± 1.4 mg/g). Correctness and independence are separate properties, and a certificate shows only one of them.
Bringing that back to the vial: even the most careful ten-vial sampling under <905> describes a lot mean and its dispersion, and the certified content of a specific vial is inferred, not measured. Any research protocol requiring a known quantity in each replicate needs either an in-lab reconstitution volume check by weight or an acceptance of the batch-level variability the sampling plan implies.
What UV-based purity misses at the fill step
The UV-transparent-excipient problem also applies at the vial-content level. Choules and colleagues at USP Biologics reported quantitative NMR analysis of commercially sourced custom synthetic peptides that found undeclared mannitol at 20% and 43% w/w — a bulking agent invisible to HPLC-UV because it is UV-transparent and poorly retained on reversed-phase columns (Choules MP et al., J Pharm Biomed Anal 178:112878, 2020, PMID 31671336). A vial certified at “99% pure” by HPLC-UV that contains 40% mannitol contains 60% peptide, not 99%. Content uniformity of the labeled active — the number a researcher actually cares about — is a different question from area-percent purity of what elutes.
The corollary for sampling: a batch COA that reports both a purity area% and an independent content mass in mg per vial contains two orthogonal pieces of information. A COA that reports only purity does not.
The distinction also affects the extinction coefficient assumption baked into every UV-detector calibration. Different amino acids absorb at 214 nm with different molar extinction coefficients — Kuipers and Gruppen measured the peptide bond at ~923 M⁻¹cm⁻¹, tryptophan at ~30× that, aromatic residues at 6× (Kuipers BJH, Gruppen H, J Agric Food Chem 55:5445-5451, 2007, PMID 17539659). Two peptides at the same UV area% therefore contain different absolute quantities per unit signal, so “purity” and “content” diverge by sequence composition even before sampling variability enters.
What our COA carries — and what it does not
Our own certificates are HPLC-UV reports with a stated content mass and a fill accuracy calculated from measured versus expected vial mass. For our BPC-157 10 mg vial (report DBAV-BPC-10-051226, tested 23 May 2026 by Testides Analytical), the analysis row records expected 10.00 mg, content 10.29 mg, purity 99.41%, fill accuracy 102.9%. That is: the sampled units averaged slightly over the labeled 10 mg, at HPLC-UV area% 99.41. The certificate carries a report number, a test date, a named third-party lab, a measured content mass and a stated method.
What it does not carry: a mass-spectrometry identity confirmation of the eluting main peak; a quantitative NMR check for UV-transparent bulking agents; a <905>-style Stage 1 or Stage 2 AV computation across ten or thirty units; or endotoxin, sterility or moisture rows. Those tests are not universal on research-grade certificates — most vendor COAs in this category are HPLC-UV only — but the honest answer to “what does this certificate actually say about the next vial” is bounded by that scope. It says the sampled composition was 99.41% at UV area% and the sampled mass was 10.29 mg. It does not certify per-vial content across the whole lot to <905> criteria, and any research protocol that depends on that level of assurance should build a weight check into its own reconstitution SOP.
Practical implications for research protocols
Three things follow, and none of them require distrusting the COA.
First, if an experimental design depends on a known absolute mass per replicate, weigh the reconstituted solution volume or the empty vial before reconstitution. A tared vial mass compared against expected fill mass takes under a minute and turns a batch inference into a per-vial measurement.
Second, if a design compares across lots, batch-mean shift and batch-variance shift are separate quantities. A different lot at the same reported purity can carry a different impurity profile, especially for the sequence-specific impurities that solid-phase synthesis generates. Documenting the lot number in the raw data is what makes that traceable later.
Third, for methods sensitive to counterion or bulking excipients, request the salt form and any excipient identity from the supplier before ordering. A UV-based COA cannot certify what it cannot see; the question has to be asked separately.
Bottom line
USP <905> gives one framework for judging content uniformity across a lot, but it is a framework for judging a sample. A COA reports what the sample yielded and — implicitly, through the sampling plan — how far that result can be extended to the rest of the batch. For lyophilized research peptides, weight variation, moisture and sequence-specific impurity distribution all vary vial-to-vial in ways a pooled batch assay averages out. The strongest COAs in this category state the method, the report number, the sampled content mass and the third-party lab; the honest reading of any of them is that they describe what the lab saw, not what the researcher will see in the specific vial opened next.
Materials described here are supplied for research use only. Not for human consumption.
References
- Choules MP, McAlpine JB, Lankin DC, McCommas JR, Fabricant DS, Lu Z, Ampofo K, Farnsworth NR, Chen SN, Pauli GF. Quantitative NMR analysis reveals significant unreported excipients in custom synthetic peptides. J Pharm Biomed Anal 178:112878, 2020. PMID 31671336.
- Kuipers BJH, Gruppen H. Prediction of molar extinction coefficients of proteins and peptides using UV absorption of the constituent amino acids at 214 nm. J Agric Food Chem 55(14):5445-5451, 2007. PMID 17539659.
- Li C, Bhavaraju S, Melanson J, Wilson DR, Sisco E, Hunter DA, Bailey CB, Aggarwal AS, Fitchette JJ, McShane KR, Faroongsarng D, Yin H, Loste D, Riggs T, Bruce V, Ryan E, Ellingson D, Bell KP, Yu LX. Multi-laboratory collaborative study to establish a fit-for-purpose HPLC method for quantitation of peptide content. J Pharm Biomed Anal 166:105-112, 2019. PMID 30640042.
- Bureau International des Poids et Mesures. Final Report on CCQM-K115.b: Peptide purity — synthetic oxytocin. 2020. Available at https://www.bipm.org/en/publications.
- U.S. Pharmacopeial Convention. General Chapter <905> Uniformity of Dosage Units. USP-NF, current revision, official 1 August 2023 (Stage 4 harmonization). Available at https://www.uspnf.com.
- U.S. Food and Drug Administration. Inspection of Lyophilization of Parenterals (7/93). Available at https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/inspection-guides/lyophilization-parenteral-793.
- U.S. Food and Drug Administration. MAPP 5019.1: Allowable Excess Volume and Labeled Vial Fill Size in Injectable Drug and Biological Products. Available at https://www.fda.gov/media/155066/download.
Research use only. Not intended for human diagnostic or therapeutic application.