Review Team (February 2016)

Bio-03 (February 2016) FOR ERP USE ONLY DO NOT DISTRIBUTE

Introduction

A single laboratory validation of the method described below has been completed and is being submitted to AOAC SPIFAN for consideration as a dispute resolution method for biotin.

Method Summary

This reverse phase HPLC method with post column protein conjugation and fluorescence detection allows for the quantitative determination of biotin in infant, pediatric, and adult nutritionals. Sample of appropriate size is mixed with 6% meta-phosphoric acid to precipitate out the protein to produce a filtrate. The filtrate is injected onto a C18 HPLC column where biotin and riboflavin are separated with a mobile phase (20% methanol in 0.02 M phosphate buffer at pH7.0). The biotin after eluting from the column binds with the streptavidin fluorescein to become a fluorescent conjugate. The conjugate is then detected by fluorescence at an excitation wavelength of 495 nm and an emission wavelength of 518 nm. A column switch is used in the method to shorten the run time from 30 min to 15 min, by eluting out riboflavin at a higher flow rate.

Single Laboratory Validation

Experimental

Precision

All fortified and unfortified SPIFAN matrices were freshly prepared and analyzed in duplicate on six days.

Accuracy

A total of eleven representative SPIFAN matrices were spiked with biotin, dissolved in 0.5% ethanol, at either 150% or 50% of the previously determined biotin level for placebos, or at either 100% or 50% of the previously determined biotin level for fortified matrices. The spiked sample was either stored at room temperature for 2 hours, or stored refrigerated for at 24 hours to allow biotin to become incorporated into the sample matrix. The spiked samples were prepared and analyzed in duplicate on 3 days. SRM 1849a with different sample sizes were prepared and analyzed. The result was compared to the CoA value.

Calibration Fit

During each analytical run 7 standards with biotin concentrations ranging from 5 to 100 ng/mL were injected before and after each sample set. Calibration curves were constructed from these standards using a polynomial regression curve (cubic-fit) and used to back calculate the concentration of each working standard in order to calculate calibration error at each level.