Low Lactose ERP - Review Book

7. Any general comments about the method?

- the method itself appears to be appropriate to address the problem at hand

- the report is not well structured, it could be imporved to make review easier to read and review. In particular it is recommeded the authors separate the actual method (to be published in OMA if accepted) from the rest of the other details not required for the method itself.

- The authors have included all the part numbers for their instruments. I highly recommend they keep such details for an internal application note, or to help their customers. But it is not needed for an AOAC method.

- recommend the authors to re-think the design of the calibration curve, as the current design does not suitably demonstrate linearity (too few points across the whole reange and too many at the low range - effectively behaving as one point). Options:-

(1) cover the whole range with equally distributed calibration points (e.g. 0.1, 5.0, 10, 15, 20, 25, 30, 35, 40 ug/kg) (2) if they feel they need calibration at very low concentration prepare a separate calibration curve including only low concentration points

- report the working range as it would be in a product (i.e. after sample preparation, etc) not according to the calibration range of the instrument

it is indicated that between 0.1 to 1 g of sample should be weighed out and diluted with 50 g of water, and further dilution may be required later. - Some more specific guidlines would be useful for the users for example specifying appropriate dilutions (and amounts to be weighed) depending on the expected lactose content of the sample

- for the in-line dialysis. When do the conditions have to be optimzed and how frequently ? (or is it possible that the conditions described in the method will be suitable for all users - in which case we don't need the details on how to optimize).

- On P1 it is written that the working range is 0.1 mg/kg to 40 mg/kg equivalent to 0.01 to 4 g/100g. Is this a typo ?

Section G(d) Inline dialysis parameters has many mistakes that need to be adressed. E.g. ...."is run with a fixed dialysis time and varying dialysis times"

Section G(d) states "controls of the setup should be run periodically". The timing needs to be better defined.

Figure 3 is not clear. Would be better if the X- axes on the two upper graphs were labelled with "dialysis time" and "transfer time" rather than "number"

Section I(a) what do the authors mean with a "validated" check standard ?

Section J(b) the legend for the equation does not match what is written in the equation

Section J(b) no need for coversion of lactose to lactose monohydrate (SMPR is for lactose)

Section K(a) the authors suggest that the usual requirment for resolution between 2 peaks is 1.0. I would suggest it is normally at least 1.5. The referenced document states that 1.0 is the minimum requirement but 1.5 is usually sought

Section K(c) what do the authors mean by "the system purity was checked.."

Section K(d) talks about F- containing matrices, and the legend for the equation also talks about concentration of fluoride ions. Clearly this is a mistake

Section K(d) the whole description of estimation of LoD / LoQ is rather confusing. But as I understand it 2 approaches have been taken. One of which results in an estimated LoQ of 0.132 g/100g of lactose monohydrate. By my calculation that is about 126 mg/100g of lactose. While the other technique results in an estimated LoQ of 8 mg/100g of lactose. The authors choose to use the LoQ of 8 mg/100g, however I would always choose the method giving me the highest value. However if they do so in this case, then the LoQ is too high for the application. I would not normally expect two estimates of LoQ to be so far apart, this needs some investigation

Section K(e) the authors have not really understood RSD(R) - this can only be assessed during an MLT, so discussions aroun RSD(R) can be reomoved

Section K(e) RSD(r) seems to have been assessed sufficiently for ML-2310 and coop milk, but not for anything else.

Section K(f) equation for estimating trueness from spike/recovery experiments is incorrect

Section K(f) can also assess trueness by comparing measured concentration vs certified value when using CRMs

V. Final Recommendation Do you recommend this method be adopted as a First Action and published in the Official Methods of Analysis of AOAC INTERNATIONAL? Please specify rationale.

No I do not recommend this method be adopted as first action.

- The calibration curve has too many points at the low concentration which effectively act as one point relative to the higher concentrationpoints. The calibration model is effectively assessed on 3 maybe 4 points.

- Insufficient data has been collected to assess RSD(r) and trueness.

- The authors should consider assessing the perfromance of the method across a broader scope of matrices.

- However I think the authors are aware of these limitations and requested more time to complete the SLV, and I recommend they do that.