A skim milk powder that passes on crude protein but fails on true protein
An incoming lot of skim milk powder is evaluated against a purchase specification of 'protein ≥ 35.0% (dry basis)'. The supplier's certificate of analysis states 36.0% protein (Kjeldahl, N × 6.38) and 4.0% moisture. Your laboratory measures moisture at 4.6% and total nitrogen at 5.68% by Kjeldahl, corresponding to 36.2% crude protein as-is. Because the figure looked high, the analyst also ran a trichloroacetic acid precipitation step before Kjeldahl and obtained 5.17% 'true protein' nitrogen, i.e. 33.0% as-is. The finished-product label declares 34 g protein per 100 g powder. A confirmatory LC-MS/MS screen for melamine returns below 0.5 mg kg⁻¹. The plant uses this powder in a recombined fermented product where protein content drives yield calculations and the label claim. Ash is 8.2% against a specification of ≤ 8.0%.
1. Convert both protein results to a dry basis and decide whether the lot meets the specification and the label claim. State which figure is the regulated 'protein' value.
Read the worked answer
Dry-basis conversion using the laboratory's own moisture of 4.6% (dry matter = 0.954):
- Crude protein (total N × 6.38): 36.2/0.954 = 37.9% dry basis
- True protein (TCA-insoluble N × 6.38): 33.0/0.954 = 34.6% dry basis
Against the specification of ≥35.0% dry basis, the lot passes on the crude figure by a wide margin and fails on the true-protein figure by 0.4 percentage points.
Against the label of 34 g per 100 g powder, the as-is crude value of 36.2% comfortably complies; the true-protein value of 33.0% does not.
Which is 'the' protein? For milk and milk products, the regulated and customary declaration is crude protein calculated as total N × 6.38. On that basis the label and the specification are both met. That does not make the question go away: the NPN fraction is nutritionally unavailable to the consumer and contributes nothing to curd formation or to the yield of the recombined fermented product, so the specification as written is not measuring the property the plant actually cares about.
Also flag the certificate of analysis: it reports 4.0% moisture while the laboratory found 4.6%. The supplier's dry-basis arithmetic is therefore built on an understated moisture, which inflates its reported protein figure. That is a supplier quality-system finding in its own right.
2. What could produce a gap of this size between crude and true protein, and which follow-up tests would discriminate between the causes?
Read the worked answer
Quantify the gap first. The difference is 5.68 − 5.17 = 0.51% N, i.e. 0.51 × 6.38 = 3.3 percentage points of apparent protein. That non-protein nitrogen is 0.51/5.68 = 9.0% of total N, against roughly 5–6% typical for fresh skim milk powder.
Candidate causes and their signatures:
- Melamine or another nitrogen-rich adulterant. Melamine is 66.7% N, so 0.5% melamine would add 0.33% N (2.1 percentage points of crude protein). The observed gap is equivalent to about 0.77% melamine - entirely excluded by the LC-MS/MS result of <0.5 mg kg⁻¹, which is five orders of magnitude below that. Rule it out.
- Added urea. Urea is 46.6% N, so about 1.1% urea in the powder would account for the gap. A specific urease/enzymatic urea assay is cheap and decisive.
- Ammonia or ammonium salts, for example from cleaning-chemical residue or from neutralisation. Measure ammonia/ammonium ion and check the ash anomaly (8.2% versus a ≤8.0% limit), which is consistent with a mineral residue.
- Proteolysis during storage. Plasmin activity in stored powder raises TCA-soluble peptides. This would be accompanied by a raised free amino acid and peptide profile, and by sensory off-notes; a casein number and SDS-PAGE profile would show it.
- Added whey or UF permeate. Permeate would raise lactose and lower the casein:whey ratio while diluting true protein. Measure lactose, total solids and the casein number.
The most informative small panel: urea, ammonia, lactose, casein number, and a stored-sample comparison against a retained reference lot of known history.
3. Given the results, what do you do with the lot, and how should the specification be rewritten?
Read the worked answer
Decision on the lot: hold it rather than release or reject outright. It meets the specification and label as written (crude protein), so a unilateral rejection is not defensible against the supplier. But three findings justify a hold: the 0.4 percentage-point true-protein shortfall, the unexplained 9% NPN fraction, and the moisture and ash discrepancies plus a CoA whose moisture does not match the measured value. Release should be conditional on the supplier explaining the NPN source; if the cause is permeate addition or undeclared dilution, reject as non-conforming to the agreed composition.
Impact on the plant: if yield models for the recombined fermented product are built on the certificate's crude protein figure, they will over-predict yield by roughly 3 percentage points of protein. Yield should be modelled on true protein or casein number, not on total N × 6.38. The NPN also changes the free nitrogen available to the starter culture, which can shift fermentation rate and flavour.
Specification rewrite:
- Specify total protein by Kjeldahl N × 6.38 as the legal/commercial basis, but add a maximum NPN limit, for example NPN ≤ 6.5% of total N, plus a minimum true protein (TCA-insoluble) figure.
- Set an explicit moisture limit with a defined reference method and state whether protein is on a dry basis.
- Require the CoA to report moisture, ash, lactose, casein number and the NPN fraction, so that a single number cannot hide a composition change.
- Add a contractual clause on adulterant screening, with melamine and urea tests at defined limits.
A supplier audit or a retained-sample comparison against earlier lots should follow if the NPN fraction cannot be explained.