Organic solvents compromise performance of internal standard (ascomycin) in proficiency testing of mass spectrometry-based assays for tacrolimus.

To the Editor: Therapeutic drug monitoring of immunosuppressive agents by liquid chromatography (LC) coupled with mass spectrometry (MS) can provide highly selective and sensitive quantification from biological fluids. Interlaboratory performance for cyclosporine, tacrolimus, and sirolimus, as assessed by LC/MS, in proficiency testing (PT) programs [e.g., College of American Pathologists (CAP) (1 ) and the International Proficiency Testing Scheme (IPTS) (2 )] has been poor, albeit improving, although individual laboratories report excellent interassay imprecision (CV 5.5%) at drug concentrations across analytical ranges (3–6). Lack of industrywide standardization may contribute to variations in PT program performance across MS laboratories because it requires that those laboratories develop “home brew” assays, use reference materials with potentially different purities obtained from numerous sources, and prepare either whole-blood calibrator/qualitycontrol (QC) materials in house or purchase commercial materials that may be verified only for a specific procedure. In addition, most PT samples are unlike routine specimens, being prepared from either artificial or whole-blood–like matrices (sometimes pooled patient blood) that are subjected to the rigors of batch preparation and shipping. I describe here an experience of specific and unusual performance of PT samples, which likely resulted from instabilities of the internal standard (IS) in organic solvents and which may be of value to scientists struggling with the development or long-term performance of MS-based methods and to PT program providers who wish to help marginally performing laboratories improve. My laboratory has single-quadrupole LC/MS and tandem LC/ MS/MS systems with independent methods for tacrolimus. Both assays use ascomycin (Sigma-Aldrich) as IS and are based on protein precipitation with zinc sulfate and organic solvent. Tacrolimus (a kind gift of Fujisawa Pharmaceutical Co., Ltd., Osaka, Japan) dissolved in acetonitrile (ACN) and stored at 40 °C in glass is used for preparing wholeblood calibrators/QC materials. A common IS stock solution in ACN is stored at 40 °C ( 1 mg in a 3-mL glass volumetric flask) and used to make the working IS for each assay. The LC/MS IS working solution is prepared by diluting the stock to 10 mL in glass with ACN (2 ng/10 L) and storing at 20 °C. For the LC/MS extraction, 10 L of IS working solution, 250 L of whole blood (0.5–40 g/L tacrolimus), 250 L of 0.1 mol/L sodium carbonate, 50 L of 300 g/L zinc sulfate heptahydrate, and 600 L of methanol are used. The LC/MS/MS IS working solution is prepared by diluting the LC/MS IS working solution to 50 mL in glass (0.2 ng/200 L), with storage as above. For the LC/MS/MS extraction, 20 L of whole blood, 80 L of 0.1 mol/L zinc sulfate heptahydrate, and 200 L of IS working solution are used. Receipt of a “poor-quality” lot of ACN in which ascomycin in the LC/ MS/MS working solution degraded within hours required temporary abandonment of that assay because the extraction is dependent on a large volume of ACN. Thus, until substitute ACN could be found and evaluated, the LC/MS method was used with the LC/MS IS working solution prepared in methanol: sufficient original stock solution prepared in “good-quality” ACN was available to do this. As controlled by daily calibration curves, QC monitoring within and across runs, and the requirement for a CV 5% for withinrun IS peak height, the LC/MS assay performed as usual with a single preparation of IS working solution. A mean deviation from targeted concentrations of 1.2% (n 33) was observed in IPTS samples, whereas QC samples in IPTS-containing runs averaged 0.3% deviation (n 30). However, after 4 months, IPTS program performance began to deteriorate until it deviated from expected values by a mean of 27.1% by 7 months (see the table in the Data Supplement that accompanies the online version of this letter at http:// www.clinchem.org/content/vol52/ issue4/), whereas QC samples maintained an acceptable bias of 1.1%. In IPTS samples only, IS peak area decreased by 50% and height by 30%. Use of the tacrolimus peak alone yielded concentrations that were either at or below the targets (as expected with no correction by IS for extraction losses or ion suppression). Fortunately, by this time 2 sources of ACN (Mallinckrodt UltimAR and EM science Omnisolv) in which ascomycin was stable over several months had been identified. The LC/ MS/MS IS working solution was therefore prepared by diluting the suspect methanolic LC/MS working solution with this new ACN. A series of IPTS samples analyzed with this solution showed a mean deviation of 10.5% (maximum, 21.1%; n 27) with QC samples at 3.4% (n 6), but all with normal IS peaks. Preparation of the LC/MS IS working solution with the goodquality ACN corrected the problem, as evidenced by analysis of a series of 18 IPTS samples (deviation, 0.8%; deviation for QC samples, 3.2%; n 6). Similarly, an LC/MS/MS IS working solution prepared from this ACN-based LC/MS working solution gave a mean deviation of 1.3% for a series of IPTS samples (n 21; deviation for QC samples, 0.5%; n 12). Similar phenomena were observed with CAP samples (data not shown). Subsequent IPTS performance of both methods has been good: 3.7% deviation by LC/MS (n 30) with 1.0% deviation for QC samples, and 2.6% (n 36) by LC/MS/MS with 2.8% deviation for QC samples. In addition to the original IS stability problem with ACN for the LC/ MS/MS assay, these observations implicated the methanolic LC/MS IS working solution as the source of the aberrant performance of IPTS samples. Interestingly, caution has been urged against preparing methanolic Letters