SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 89, NO. 5,

Transcription

1 SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 89, NO. 5, FOOD CHEMICAL CONTAMINANTS Interlortory Study of the Chrm ROSA Sfe Level Afltoxin M1 Quntittive Lterl Flow Test for Rw Bovine Milk ROBERT SALTER, DAVID DOUGLAS, MARK TESS, BOB MARKOVSKY, nd STEVEN J. SAUL Chrm Sciences, Inc., 659 Andover St, Lwrence, MA Collortors: F. Brcellos; A. Brndt; K. Chllquist; G. Dixon; C. Fierro; A. Gillelnd; P. Gutierrez; C. Hll; T. Hickerson; K. Jones; K. Keys; M. Kingsley; D. Klein; R. Lm; C. Lucero; B. Mlone; D. Mnor; K. McClrn; R. McDniel; J. Miller; T. Moerly; J. Nndre; A. Ree; B. Shffer; P. Stetzel; K. Swcin; T. Tieso; N. Turnge; M. Willims; nd S. Wolff An interlortory study of 21 pulic helth, stte griculture, nd industry lortories in the United Sttes tested rw commingled ovine milk contining fltoxin M1 using the Chrm Rpid One Step Assy (ROSA) Sfe Level Afltoxin M1 Quntittive lterl flow method. Blind coded smple pirs were fortified with 0, 300, 350, 400, 450, 500, nd 550 prts per trillion (ppt) fltoxin M1. A ROSA reder quntittively interpreted test strips with ppt redings. Redings 400 ppt were interpreted s negtive, nd redings >400 ppt were interpreted s positive. Initil positive smples were susequently ssyed 2 dditionl times. If oth retest results were >400 ppt, the smple ws clled positive/ ctionle reltive to U.S. nd Codex levels, 500 ppt. The concentrtion of 400 ppt ws chosen for the positive/negtive interprettion to provide 90% sensitivity with 95% confidence t the 500 ppt legisltive level. The comined flse negtive rte ws <5% (4 of 83) for smples t 500 nd 550 ppt. The flse violtives t 0, 300, 350, 400, nd 450 ppt (n = 42 t ech level) were 0, 0, 21, 14, nd 93%, respectively. The 90% positive concentrtion with 95% confidence ws 503 ppt y proit nlysis. The verge intrlortory repetility ws 11% nd verge interlortory reproduciility ws 13% for the fortified smple pirs. High-performnce liquid chromtogrphy nlysis of the study smples y 5 lortories showed 38% flse negtives with the 500 nd 550 ppt smples, nd 0% flse-violtive rte with smples less thn the 500 ppt ction level. The U.S. nd Codex estlished ction level for fltoxin M1 in milk is 500 prts per trillion (ppt; 1, 2). Rpid screening methods such s enzyme-linked immunosorent ssy, immunoffinity, nd lterl flow tests Received Jnury 25, Accepted y AP Mrch 31, Corresponding uthor's e-mil: os@chrm.com re used y industry nd stte lortories for screening milk smples (3, 4). Positive smples my require further nlysis y vlidted methods such s the officilly pproved high-performnce liquid chromtogrphy (HPLC) methods for fltoxin M1 in milk (5 7). With ny methodology, there re concerns out the sensitivity, precision, nd reproduciility of the method nd the susequent rte of flse-positive, flse-violtive (positive test result with non-ctionle levels in the smple), nd flse-negtive results (8). Rpid screening methods need to provide detection t the ction level ut not e overly sensitive s to cuse the loss of milk due to flse violtives (9). The Chrm Sfe Level Afltoxin M1 Quntittive (SLAFMQ) test is colloidl gold lterl flow immunossy. Afltoxin M1 in milk smple competes with the ntiody gold eds for inding to 2 test lines. Remining unound inder forms on the control line. The test nd control lines re compred with reflectnce reder, nd ppt concentrtion is determined with n lgorithm. A negtive interprettion with reding of 400 ppt nd positive interprettion with reding >400 ppt ws designed to detect 500 ppt, the U.S. nd Codex violtive level t 90% positive with 95% confidence. Retesting n initil positive smple 2 dditionl times ws confirmtion step to reduce flse-violtive smples. The purpose of this interlortory study ws to determine the multilortory performnce chrcteristics of the SLAFMQ test in rw milk nd to compre these results with vlidted HPLC methods. This study ws orgnized y regionl stte helth lortories nd Chrm Sciences. Dt were sent to n independent third prty who sent results to Chrm Sciences for computtion. Methods SLAFMQ Procedure Equipment nd regents were supplied to volunteer lortories long with 4 prestudy smples t 0, 300, 400, nd 500 ppt to fmilirize the lortory nlysts with the SLAFMQ lterl flow method. Rpid One Step Assy (ROSA) Reder v ws used for nlysis. The SLAFMQ method is s follows: (1) Pipet 300 L cold (0 7 C) dilution uffer into microcentrifuge tue. (2) Mix

2 1328 SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 89, NO. 5, 2006 Tle 1. Testing scenrios nd interprettion of test results Initil test result Retest No. 1 result Retest No. 2 result Interprettion 400 ppt or less negtive NA NA Negtive/non-ctionle 401 ppt or greter positive 400 ppt or less negtive 400 ppt or less negtive Negtive/non-ctionle 401 ppt or greter positive 400 ppt or less negtive 401 ppt or greter positive Negtive/non-ctionle 401 ppt or greter positive 401 ppt or greter positive 400 ppt or less negtive Negtive/non-ctionle 401 ppt or greter positive 401 ppt or greter positive 401 ppt or greter positive Positive/ctionle, 500 ppt or greter NA = Not pplicle to retest n initil test result 400 ppt or less. Positive interprettion y ROSA reder. rw milk, nd pipet 300 L into the microcentrifuge tue. Cp nd shke mixture vigorously for 5 s. Keep cold t 0 7 C. (3) Plce SLAFMQ strip in 45 2 C incutor, peel open smple comprtment, pipet 300 L of the ove mixture into comprtment, resel strip, nd close incutor lid, which strts n 8 min timer. (4) After 8 min, remove SLAFMQ strip nd visully inspect control line for even development to determine tht the test result is vlid. (5) Insert vlid strip into ROSA reder for 5 s nlysis. The reder displys the determined ppt concentrtion nd interprets vlue >400 ppt s positive nd vlue 400 ppt s negtive. For the study, testers retested ny positive smple 2 more times. If oth the dditionl tests were positive, the smple ws positive/ctionle. If either retest ws negtive, the smple ws negtive/non-ctionle. Blind Study Smple Preprtion Two rw milk smples from frm ulk tnks were nlyzed to contin <50 ppt fltoxin M1 using lterl flow test, Mximum Residue Level for Afltoxin M1 method (LF- MRLAFM, Chrm Sciences), to screen Europen Union mximum residue levels of fltoxin M1, 50 ppt. Sigm fltoxin M1 stndrd (Ct. No U) ws mde to stock concentrtion of 5 g fltoxin M1/mL (in cetonitrile) nd verified y Vrin spectrophotometer t A 350nm = Qulified rw milk ws prepred to contin 300, 400, nd 500 ( 3% sed on volumetric nd pipet error) ppt from the fltoxin M1 stndrd. Another qulified rw milk smple ws prepred to contin 0, 350, 450, nd 550 ( 3%) ppt. Milk smples in 5 ml portions were seled under nitrogen in glss vils. Duplicte smples were lind coded, shipped on ice, nd tested y prticipnts within 1 week. Four lortories with Americn Oil Chemist proficiency certifiction or FDA-CFSAN certifiction nd one noncertified lortory nlyzed smples y AOAC HPLC methods. Sttisticl Design The design of the study ws sed on interntionl interlortory study protocol ISO-5525 (10). All lortories tested ech smple once nd reported these initil results. Dt nlysis for outliers nd sttisticl prmeters for repetility (r) nd reproduciility (R) were clculted with initil results of the lind coded smple pirs ccording to ISO Smples with n initil positive result were ssyed 2 more times, nd smples were interpreted positive/ctionle or negtive/non-ctionle s descried in SLAFMQ procedure. The 90% positive concentrtions with 95% confidence level were determined from dose-response versus concentrtion curves. The SLAFMQ method positives nd initil test positives, per totl numer of smples tested t ech concentrtion, were nlyzed y XL-Stt proit nlysis with the 95% confidence vlue converted to one-til (11). The Ntionl Conference of Interstte Milk Shipments (NCIMS) hs used similr sttistics to vlidte tht ntiiotic screening tests hve 90% positive concentrtions with 95% confidence t or elow U.S. sfe levels/tolernces (12). To mintin proper lind study protocol, the dt from ech prticipnt were forwrded to the Texs Deprtment of Stte Helth Services, n independent third prty tht ws not involved in smple preprtion, for collection, scoring, nd decoding. Results nd Discussion The SLAFMQ method testing scenrios for the initil test result nd susequent retests, nd the interprettion of these results in terms of negtive/non-ctionle nd positive/ctionle, re shown in Tle 1. Only smple tht tested positive on the initil test nd positive on the 2 susequent retests ws considered to e t 500 ppt or greter nd interpreted to e positive/ctionle smple. The initil test results in ppt for ll 294 smples nlyzed re presented in Tle 2 nd in Figure 1. The smples tht tested >400 ppt on the initil test were retested 2 dditionl times, nd these results re presented in Tle 3. The interprettion of ll smples from dt from Tles 2 nd 3 in terms of negtive/non-ctionle nd positive/ctionle is presented in Tle 4. Smples tht were positive t 350, 400, nd 450 ppt cn e considered flse-violtive smples since they were found positive/ctionle y the method ut contined fltoxin M1 less thn the 500 ppt violtion/ction level (9). Outliers were determined y Cochrn nd Grus nlysis with 1 Cochrn outlier found in ech of the 0 ppt smples nd

3 SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 89, NO. 5, Tle 2. Initil reding of lind coded smples using chrm SLAFMQ ssy Lortory ID Added concn, p pt c c d d Vlues 400 nd lower with either 500 or 550 ppt smples re flse negtive. Vlues 401 nd greter with smples contining <500 ppt re initil test flse-violtive (positive results with smple s contining non-ctionle levels of fltoxin M1). Reding results greter thn the limit (400 ppt). c Cochrn outlier excluded from nlysis. Cochrn strggler not excluded from nlysis. d

4 1330 SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 89, NO. 5, 2006 Figure 1. Initil lortory determintions of ech dded concentrtion nd the clculted mens. Negtive ( ), 300 ppt ( ), 350 ppt ( ), 400 ppt (x), 450 ppt ( ), 500 ppt ( ), nd 550 ppt (+) smples were plotted versus their spiked concentrtions. Verticl lines m0, 300, 350, 400, 450, 500, nd 550 mrk the men determintions of the 21 lortories for ech of the fortified concentrtions. Redings greter thn the 400 ppt control point ( ) were initil test positive. 550 ppt smples (10). All results t the 0 nd 300 ppt were negtive/non-ctionle y the SLAFMQ method. At 350 ppt, the 9 flse violtives represented 21% of the totl, while t 400 ppt the 6 flse violtives represented 14% of the smples. At 500 ppt, there were 3 of 42 (7%) flse-negtive/non-ctionle results, nd t 550 ppt there ws 1 of 41 (2%) flse-negtive/non-ctionle results. The SLAFMQ method dose response, the initil test result dose response, nd one dt clculted 90% positive concentrtions with 95% confidence re presented in Tle 5. The control point t 400 ppt nd retesting procedure ws selected to provide 90% positive concentrtion with 95% confidence t the 500 ppt ction level while limiting flse-violtive results. The SLAFMQ method 90% positive concentrtion with 95% confidence ws 503 ppt, which is shift from 470 ppt clculted from the initil test result. The SLAFMQ method s compred to running only the initil test reduced the flse-violtive results t 350 ppt from 52 to 21% nd t 400 ppt from 50 to 14%. The flse-violtive rte from ll smples t 400 ppt nd less ws 15 positives out of 167 smples, or 9%. The flse-violtive rte t 450 ppt ws 93%. A high flse-violtive rte t 450 ppt ws expected, s the SLAFMQ threshold level of 400 ppt ws chosen to minimize flse negtives. Acceptnce of flse-violtive results from smples contining >400 ut <500 ppt ws consistent with recent single lortory evlution of fltoxin M1 tests (Trujillo et l., poster presented t the 2005 AOAC Annul Meeting in Orlndo, FL). The flse-negtive rte for 500 nd 550 ppt smples ws 4 negtives out of 83 results, or 4.8% of the smples. The flse-negtive rte for the SLAFMQ method ws twice tht of performing just the initil test where the flse-negtive rte ws 2.4%. The flse-negtive rte of the initil test met criteri for single lortory evlutions of NCIMS screening tests (12). Performnce criteri for multiple lortory testing hve not een estlished y NCIMS. Multiple lortory evlutions of methods typiclly follow AOAC nd ISO/IDF guidelines nd re more roust thn single lortory evlutions. The HPLC results from 5 lortories re reported in Tle 6. There were no flse-violtive results in the finl reported determintions. A 38% flse-negtive rte ws found since 9 of 24 smples t 500 or 550 ppt fltoxin M1 were reported to contin less thn the 500 ppt ction level. HPLC results indicted some interlortory vrition with Lortory A, showing close greement to the prepred study stndrds. HPLC methods do not pply threshold for vrition t 500 ppt to chieve 90% positive concentrtion with 95% confidence. Lortories A, C, D, nd E performed AOAC Method (5), which yielded more cceptle results thn dt from performing Method (6) y Lortory B. Overll, the HPLC determined concentrtions of fltoxin M1 in the smples were consistent with the prepred fortified fltoxin M1 concentrtions in milk. The intrlortory mens t ech concentrtion ( m 0 m 550 ) from performing the SLAFMQ method re presented in Tle 7 nd Figure 1. The men vlues from nlysis of the 300, 400, nd 500 ppt smples correlted within 3% of the prepred concentrtions. The men vlues from nlysis of the 350, 450, nd 550 ppt smples correlted within 14% nd trended more positive thn the prepred concentrtions. This positive is when using this rw milk to prepre smples contining 350, 450, nd 550 ppt fltoxin M1 my explin the positive trend in these smples nd why the 450 ppt smple men ( m 450 = 505 ppt) ws greter thn the 500 ppt smple men ( m 500 = 495 ppt). Intrlortory repetility nd interlortory reproduciility of dt from nlysis of the lind smple pirs (Tle 2) were clculted nd re presented in Tle 7. Repetility is the rnge of determintions tht cn e expected from multiple nlyses on the sme smple in single lortory. Reproduciility is the rnge of determintions tht could e expected from multiple nlyses of identicl smples in multiple lortories. The CV r % of repetility (RSD r ) t ech concentrtion ws <16% nd on verge ws 11% of the determined concentrtion. The CV R % of reproduciility (RSD R ) t ech concentrtion ws <20% nd on verge ws 13% of the determined concentrtion. The repetility (r) nd reproduciility (R) vlues represent 2.8 stndrd devitions of the lortory determintions. The vlues t 550 ppt, of r = 101 nd R = 126, were hlf of those determined y the HPLC method t 580 ppt, which were r = 203 nd R = 310 (13). These sttisticl prmeters represent the 95% vrition rnge expected from identicl smple determintions within lortory (r) nd etween different

5 SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 89, NO. 5, Tle 3. Duplicte retests of initil positive smples Lortory ID Added concn, ppt c c d c e c e e e c d e The 300, 350, 400, nd 450 ppt smples testing >400 ppt on oth retests were SLAFMQ flse violtives. Reding results greter thn the limit (400 ppt). Initil flse negtive; result not retested. Initil test result ws n outlier nd excluded. The 500 or 550 ppt smples tht tested negtive nd were flse negtive. lortories (R). Lower r nd R vlues nd very low HorRt vlues, HorRt r < 0.3 nd HorRt R <0.5, indicte tht the SLAFMQ method my hve greter precision thn HPLC for quntittion of fltoxin M1 levels in milk. Conclusions The SLAFMQ ssy for rw milk in 21 lortory interlortory study detected U.S. nd Codex ction levels with 90% positive concentrtion with 95% confidence t 503 ppt nd 4.8% flse-negtive rte. Flse violtives were minimized using confirmtion procedure tht required smples with initil positive results to e retested twice, nd for oth retests to e positive. The clculted repetility (r) nd reproduciility (R) for the SLAFMQ method were lower thn pulished vlues for HPLC methods t comprle concentrtions. The SLAFMQ method hd greter confidence t detecting ctionle smples, t 500 nd 550 ppt, thn HPLC methods performed on the sme smples.

6 1332 SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 89, NO. 5, 2006 Tle 4. Positive/ctionle or negtive/non-ctionle determintions fter retesting Lortory ID Added concn, ppt = Positive/ctionle results. = Negtive/non-ctionle results. Smples t 500 nd 550 ppt tht tested negtive were flse negtives. The 300, 350, 400, nd 450 ppt smples testing positive were flse violtives. Initil test result ws n outlier nd excluded. Tle 5. Dose responses of SLAFMQ method nd initil test with 90% positive concentrtions with 95% confidence SLAFMQ method Initil test Added concn, ppt No. of smples tested No. positive % Positive No. positive % Positive % Positive concentrtion with 95% confidence, ppt Person Chi squre Outliers t 0 nd 550 removed. Acknowledgments We thnk Gene Wright of the Deprtment of Stte nd Helth Service (DSHS), Austin, TX, for his ssistnce in collecting, scoring, nd decoding dt. We lso thnk the following collortors for their prticiption: C. Hll nd M. Kingsley, Texs DSHS, Austin, TX G. Dixon nd N. Turnge, Trrnt County Pulic Helth, Fort Worth, TX K. McClrn nd K. Swcin, City Luock Pulic Helth, Luock, TX M. Willims nd S. Wolff, Sn Antonio Metro Helth Deprtment, Sn Antonio, TX

7 SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 89, NO. 5, Tle 6. HPLC determintions of smples reported in ppt HPLC results reported y ls HPLC rw dt of L C (run 1) nd L B (runs 1 3) Smple No. Added concn, C 1 ppt A B c C 1 d C 2 d D e E e reported d B (1st run) c B (2nd run) c B (3rd run) c f f f 520 f 598 f 457 g 460 g 476 g g 270 g 850 f f 500 f 584 f 363 g 470 g 514 f 794 f 460 g 310 g 720 f f f f 400 g 474 g 307 g 520 f 418 g 645 f 110 g 380 g 710 f f ND f 540 f 609 f 535 f 560 f 559 f 828 f 340 g 510 f 760 f 28 0 ND c d e f g HPLC vlues reported y lortories were in pp nd were converted to ppt. The 450, 400, 350, nd 300 ppt smples testing >500 ppt were flse-violtive results. Lortory A performed AOAC Method , Affinity purifiction of fltoxin M1 followed y derivtiztion nd HPLC-fluorescence detection. Lortory B performed (HPLC) AOAC Method nd reported possile prolems with injector pprtus nd requested verge of 3 determintions e reported. Rw dt re displyed s 1st, 2nd, nd 3rd run (lst 3 columns). Lortory C performed (HPLC) AOAC Method nd initilly reported vlues C 1 reported. When sked to investigte positive is, Lortory C found concentrtion prolem with the clirtion stndrd nd corrected the originl dt to C 1. The lortory lso retested the smples 1 week lter fter pplying correct stndrd, results C 2. Lortories D nd E performed AOAC Method modified for reduced volume smple nd tested smples 1 14 tht were converted to the HPLC smple lind codes, Results greter thn the ction level (500 ppt); true positives. The 500 or 550 ppt smples tht tested lower thn the ction level nd were flse negtives. Tle 7. Chrm SLAFMQ results showing men, repetility, reproduciility, nd HorRt sttistics from dt in Tle 2 Intrlortory repetility sttistics Interlortory reproduciility sttistics Added concn, ppt Men ( m) of SLAFMQ determintions StDev r CV r % (RSV r ) HorRt r vlue Repetility (RSV r / CV R % r = 2.8 (StDev r ) PRSV R ) SD R (RSV R ) Reproduciility R = 2.8 (SDR R ) HorRt R vlue (RSV R / PRSV R )

8 1334 SALTER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 89, NO. 5, 2006 C. Lucero nd C. Fierro, El Pso City Env. Helth District, El Pso, TX K. Keys, Northest Texs Pulic Helth District, Tyler, TX A. Brndt, Pioneer Ls, Artesi, NM P. Gutierrez nd A. Ree, New Mexico Deprtment of Agriculture, Ls Cruces, NM K. Jones nd R. McDniel, North Crolin Deprtment of Agriculture Food nd Drug Protection, Rleigh, NC T. Tieso, Nersk Deprtment of Agriculture, Lincon, NE K. Chllquist, Knss Deprtment of Agriculture, Topek, KS P. Stetzel, Iow Deprtment of Agriculture, Des Moines, IA A. Gillelnd, Georgi Deprtment of Agriculture, Atlnt, GA D. Klein, Arizon Deprtment of Agriculture, Phoenix, AZ D. Mnor, Arknss Deprtment of Agriculture, Little Rock, AR F. Brcellos, Oklhom Deprtment of Agriculture, Oklhom City, OK B. Mlone, Trilogy Lortory, Wshington, MO B. Shffer, DFA, Knss City, MO T. Hickerson nd T. Moerly, DFA, Grpevine, TX R. Lm nd J. Miller, Florid Deprtment of Agriculture, Tllhssee, FL J. Nndre, U.S. Food nd Drug Administrtion, Knss City District Lortory, Lenex, KS References (1) Office of Regultory Affirs, Section CPG , 11/29/05, cpgfod/cpg html (2) Codex Alimentrius Commission (2001) Stndrd Afltoxin Levels in Milk, downlod/stndrds/399/cxs_232e.pdf (3) Littlefield, J. (1998) Memo: Alert of Possile Afltoxin Contmintion in Some Milk Supplies, Texs Deprtment of Helth, Austin, TX (4) Rowlnd, R. (2005) Memo: Afltoxin Screening, Iow Deprtment of Agriculture nd Lnd Stewrdship, Des Moines, IA (5) Officil Methods of Anlysis (2005) 18th Ed., AOAC INTERNATIONAL, Githersurg, MD, Method (6) Officil Methods of Anlysis (2005) 18th Ed., AOAC INTERNATIONAL, Githersurg, MD, Method (7) Officil Methods of Anlysis (2005) 18th Ed., AOAC INTERNATIONAL, Githersurg, MD, Method (8) Henry, S.H., Whitker, T., Rni, I., Bowers, J., Prk, D., Price, W., Bosch, F.X., Pennington, J., Verger, P., Yoshizw, T., vn Egmond, H., Jonker, M.A., & Coker, R. (June 2001) Afltoxin M1, Section 31, Screening Tests, Proceedings of JECFA, jecf/jecmono/v47je02.htm#3.1 (9) Chrm, S.E. (1994) Diry Food Environ. Snit. 14, (10) ISO (1994) Accurcy Trueness nd Precision of Mesurement Methods nd Results, Interntionl Stndrd Orgniztion, Brussels, Belgium (11) XL-Stt, Addinsoft, New York, NY, (12) U.S. Food nd Drug Administrtion, Center for Veterinry Medicine (2002) Dt Requirements for Milk Screening Tests Leled for Testing Milk Tnkers t the Bulk/Tnk Tnker Truck for Drug Residues Tests with Instrument Reders/Printers Only, Rockville, MD (13) Tuinstr, L.G.M.T., Roos, A.H., & vn Trijp, J.M.P. (1993) J. AOAC Int. 76,