Evaluation of the Hologic Gen-Probe PANTHER, APTIMA Combo 2 Assay in a Tertiary Care Teaching Hospital

Transcription

1 AJCP / Originl Article Evlution of the Hologic Gen-Probe PANTHER, APTIMA Combo 2 Assy in Tertiry Cre Teching Hospitl Annie Cheng, MT, nd Jmes E. Kirby, MD From the Deprtment of Pthology, Beth Isrel Deconess Medicl Center, Boston, MA. Key Words: Chlmydi trchomtis; Neisseri gonorrhoee; Nucleic cid mplifiction test; Sexully trnsmitted diseses DOI: /AJCPFQ25SQVZAWHZ ABSTRACT Objectives: To evlute the performnce of the Hologic Gen-Probe (Sn Diego, CA) PANTHER system. Methods: The performnce of PANTHER ws compred with the Hologic Gen-Probe TIGRIS nd/or Roche (Indinpolis, IN) COBAS AMPLICOR systems through testing of ptient specimens nd the spiked-urine mtrix. Results: After discrepnt resolution, PANTHER demonstrted 99.3% (95% confidence intervl [CI], 96.0%-99.9%) positive nd 100% (98.5%-100.0%) negtive greement for Chlmydi trchomtis (CT) nd 100% (96.6%-100.0%) positive nd 100% (98.6%-100.0%) negtive greement for Neisseri gonorrhoee (NG) for ll mle, femle, unsexed, nd NG-spiked femle urine specimens combined. For other specimen types collectively, the PANTHER demonstrted 100% (95% CI, 90.6%-100.0%) positive nd 100% (88.3%-100.0%) negtive greement for CT nd 90.9% (62.8%-98.4%) positive nd 100% (93.5% %) negtive greement for NG. Anlyticl sensitivity of the PANTHER in urine mtrix ws similr to the TIGRIS system. Conclusions: The PANTHER system provides n excellent new ddition to options for detecting CT nd NG, is pproprite for testing urine smples, nd will fcilitte high-throughput testing in the clinicl lbortory. The prevlence of Chlmydi trchomtis (CT) nd Neisseri gonorrhoee (NG) infection continues to increse or remins high round the world. 1-3 Therefore, the US Preventive Services Tsk Force, 4 Centers for Disese Control nd Prevention, 5 nd professionl orgniztions 6 hve dvocted n intensive screening effort to reduce disese burden nd prevent the sequele of untreted infection, such s pelvic inflmmtory disese nd ssocited infertility. For exmple, recent recommendtions in the United Sttes include screening ll sexully ctive women younger thn 25 yers for CT. 4 Rpid, sensitive, nd ccurte nucleic cid mplifiction testing methods re needed to respond to this public helth mndte. Severl CT/NG nucleic cid mplifiction testing pltforms hve been introduced over the pst decde. However, some ssys hve shown suboptiml performnce chrcteristics (ie, cross-rectivity with nongonococcl Neisseri species). This is of specil concern in lower-prevlence screening popultion becuse of the consequent low positive predictive vlue of these tests. 7 Recently, the Hologic Gen-Probe (Sn Diego, CA) APTIMA Combo 2 (AC2; PANTHER system) ws clered by the US Food nd Drug Administrtion (FDA) in the United Sttes for the detection of CT nd NG in vginl nd endocervicl swbs, mle urethrl swbs, nd ThinPrep PreservCyt (Hologic) cervicl cytology specimens. To our knowledge, there hs been only one previously published study exmining the performnce of the AC2 ssy on the PANTHER system, limited to exmintion of the nlyticl specificity for NG. 8 However, extensive literture supports the robust performnce of the AC2 ssy run in mnul, semiutomted, nd fully utomted TIGRIS DTS systems (Hologic Gen- Probe). 9,10 Therefore, here we sought to verify the clinicl Am J Clin Pthol 2014;141: DOI: /AJCPFQ25SQVZAWHZ 397

2 Cheng nd Kirby / Evlution of the Hologic PANTHER System performnce chrcteristics of the Hologic Gen-Probe AC2 (PANTHER system) ssy (herefter referred to s PAN- THER) nd to determine whether performnce chrcteristics of the AC2 were ffected by the pltform on which it ws run. Hence, specific comprison ws mde to performnce on the TIGRIS DTS pltform. Additionl comprisons were lso mde for subset of ptients who were initilly tested for CT/NG using the Roche (Indinpolis, IN) COBAS AMPLI- COR system (COBAS) nd for whom ll NG+ results were confirmed by dditionl testing. Notbly, the PANTHER ws not clered by the FDA for testing femle nd mle urine specimens. Therefore, we lso sought to vlidte off-lbel use of urine specimens on the PANTHER to expnd the smple types vilble for outptient screening. Mterils nd Methods Chrcteristics of Clinicl Smples nd Testing Methods The described nlysis ws performed s prt of norml qulity ssurnce ctivities relted to verifiction nd vlidtion of methods being considered for doption in our clinicl lbortory prctice. The underlying intent of these studies ws not to generte generlizble knowledge. As such, institutionl review bord pprovl ws not required. Five neighboring institutions (Brighm nd Women s Hospitl, Cmbridge Hospitl, Boston Medicl Center, Msschusetts Generl Hospitl, nd Needhm Hrvrd Vngurd) supplied us with deidentified coded smples tht were previously tested nd scored s positive or negtive for CT nd NG using the TIGRIS. Smple types included mle urethrl nd femle endocervicl swbs, mle nd femle urine smples, nd smll number of ThinPrep PreservCyt smples. Urine smples from one institution were not differentited s to femle or mle sex nd therefore re referred to s unsexed urine specimens. Notbly, the PANTHER requires use of the sme smple trnsport medi s the TIGRIS. Therefore, ll smples previously collected for the TIGRIS were run on the PAN- THER without modifiction. Smples were preselected to enrich for NG+ nd CT+ specimens to evlute positive greement with greter power. Both the PANTHER nd TIGRIS use the AC2 ssy, which is bsed on trnscriptionmedited mplifiction of species-specific trgets in the 23S ribosoml RNA (rrna) of CT nd 16S rrna of NG. Specific mplifiction of these trgets leds to distinct luminescent signls, which re quntified s reltive light units (RLUs) nd used to ssign smples to negtive, equivocl, nd positive ctegories. Smples were kept frozen t 20 C fter initil testing nd thwed in our lbortory immeditely before testing in this study. We lso tested dditionl mle nd femle urine smples tht were previously nlyzed in our lbortory using the COBAS. These smples were stored t 20 C in their originl sttes (ie, without ddition of trnsport medi with nucleic cid stbilizers per the FDA-clered collection protocol for the COBAS) nd preselected for nlysis bsed on prior results (ie, to enrich for CT+/NG+ specimens). In contrst to the AC2, the COBAS uses conventionl end-point polymerse chin rection (PCR) to detect the multicopy cryptic plsmid from CT nd the single-copy M.NgoPII gene from NG. 11 Becuse of known COBAS cross-rectivity issues, ny COBAS NG+ smple ws scored s true positive only fter confirmtion using the Light Cycler (LC) bsed (Roche), rel-time PCR ssys specific for the NG 16S rrna nd PAP genes. 7 We previously determined tht these confirmtory ssys were more sensitive thn the COBAS ssy 7 nd therefore cn relibly conclude tht COBAS NG+ confirmtory test negtive smples re true negtives. Prior to retesting by the PANTHER, these smples were thwed to room temperture nd vortexed for even resuspension. Then, 3 ml ws trnsferred to n APTIMA urine trnsport tube. Subsequent to PANTHER testing, these smples were retested on the TIGRIS for comprtive evlution nd to verify coding by the outside lbortories. Thus, results from in-house collected urines were compred cross three ssy pltforms (COBAS, TIGRIS, nd PANTHER). Accurcy Of the 464 smples evluted, 395 were urine, 54 were swbs, nd 15 were ThinPrep PreservCyt smples. Among these smples, results of nine urine nd three swb smples ultimtely were not included due to unresolvble discrepncies, s explined in the text. Our gol ws to test t lest 50 CT+, 50 NG+, nd 100 CT /NG mle nd femle urine smples per recommended guidelines in Cumitech However, becuse of the low prevlence of NG in our ptient popultion, we were ble to obtin only smll number of NG+ femle urine smples. To compenste for this limited number, we performed spiking experiments in femle urine mtrix. Discrepnt Resolution Any discrepnt smple initilly tested on the TIGRIS ws tested gin on both the PANTHER nd TIGRIS. Any discrepnt smple initilly tested on the COBAS ws tested gin on the COBAS, PANTHER, nd TIGRIS if sufficient specimen were vilble. In ddition, NG discrepnt smples were lso retested using the LC confirmtory ssys described bove. Discrepncies were resolved bsed on greement chieved on retesting, s discussed in the Results section, tking into ccount the potentil for specimen degrdtion in unpreserved specimens. Smples with 398 Am J Clin Pthol 2014;141: DOI: /AJCPFQ25SQVZAWHZ

3 AJCP / Originl Article n unresolved discrepncy were excluded from subsequent nlysis for both CT nd NG nlytes. Spiking Studies Becuse of the low prevlence of NG in femle ptients in the greter Boston vicinity, dditionl nlysis of potentil femle urine mtrix effects ws done through the use of spiking studies of 47 COBAS CT nd one COBAS CT+ femle urine specimens. In totl, 2 ml of urine from ech smple ws trnsferred into n APTIMA urine trnsport tube nd screened for CT nd NG on the PANTHER. A second set of 2-mL liquots ws spiked with 100 ml of fresh suspension (500 colony-forming units [CFU]/mL bsed on McFrlnd conversion nd subsequently confirmed by CFU counts from suspension dilutions) mde from n overnight (24-hour) NG (43069, Americn Type Culture Collection [ATCC], Mnsss, VA) gr culture. Ech of these NG-spiked femle urine specimens ws then mixed with 2 ml of APTIMA trnsport medi for finl concentrtion of 12.5 CFU/mL of NG per smple nd tested on the PANTHER. Anlyticl Specificity A series of ATCC qulity control strins nd clinicl isoltes, identified bsed on phenotypic chrcteristics or biochemicl identifiction by the Vitek 2 instrument (Bio- Mérieux, Durhm, NC), were grown on chocolte or sheep blood gr for 24 hours, resuspended in sterile sline to 0.5 McFrlnd, nd diluted in pooled CT /NG urine to give n pproximte concentrtion of CFU/mL. This concentrtion, found in urinry trct infections or in significntly contminted urine specimens, ws deemed relevnt for exmining nlyticl specificity. Then, 3 ml of ech spiked urine specimen ws trnsferred into APTIMA urine trnsport tubes nd evluted on the PANTHER. Reltive Anlyticl Sensitivity in the Urine Mtrix Limiting dilution studies of the AmpliPROBE CT/GC control mteril (Bio-Rd, Hercules, CA) were used to compre the reltive nlyticl sensitivity of the PANTHER nd TIGRIS. The AmpliPROBE regent contins stbilized mixture of CT LGV type II strin 434 elementry bodies nd whole-cell lyste of NG. 13 However, the bsolute concentrtion of CT nd NG is not provided by the mnufcturer. Therefore, reltive but not bsolute limit of detection (LOD) ws determined. To ppropritely simulte ptient urine smples nd ensure ccurte dilution of control mteril cross multiple replictes nd multiple dilutions, we prepred test smples with two components consisting of urine nd control mteril stbilized in urine trnsport buffer. Notbly, the control RNA trget is intrinsiclly unstble in urine. Therefore, control mteril ws diluted in trnsport medi first nd then mixed with urine rther thn the converse. This is similr to the procedure used by the mnufcturer in limiting dilution studies described in the PANTHER 14 nd TIGRIS 15 pckge inserts. Specificlly, to perform this nlysis, series of 10-fold dilutions of the AmpliPROBE control mteril were mde in APTIMA urine trnsport medi, rnging from 50- to fold dilutions. Then, 2 ml of ech dilution ws trnsferred to n empty, lbeled APTIMA urine trnsport tube. Four tubes were prepred from ech dilution, nd these were used for initil screening. Thirty different ptient urine smples tht hd previously been confirmed negtive for both CT nd NG were pooled nd used s the urine smple mtrix. Of this pooled urine, 3 ml ws then dded to ech smple tube contining specific dilutions of the CT/NG control mteril nd mixed by gentle inversion. After the initil qudruplicte screening, replicte screening ws expnded to llow sttisticl comprison tht ws resonbly powered. Seril dilutions t, below, nd bove the lowest concentrtion (highest dilution) tht exhibited rectivity for CT nd NG were prepred, now in replictes of 20. These replicte dilutions (rnging from fold to fold) were then tested in prllel on the PANTHER nd TIGRIS to determine the reltive LOD. Sttisticl Anlysis Positive, negtive, nd overll percent greement nd 95% confidence intervls (listed s rnge in prentheses following percent greement) were clculted s recommended in Clinicl nd Lbortory Stndrds Institute s guidelines. 16 Percent greement ws determined preferentilly insted of clinicl sensitivity nd specificity becuse of the study design, which ws comprison study not referenced to ptient infectious sttus nd n nlyticl gold stndrd. 16 LOD ws determined by probit nlysis s previously described 17 using SPSS version 18.0 (SPSS, Chicgo, IL). For the most ccurte determintion of the LOD, only dt below the 100% positivity level were used for extrpoltion of the LOD. 17 Results The PANTHER pltform ws evluted for ccurcy nd precision. A mjor gol ws to verify the off-lbel use of the PANTHER for detection of CT nd NG in mle nd femle urine specimens. Therefore, more extensive nlysis ws performed for these smples types, including determintion of reltive nlyticl sensitivity in comprison with the TIGRIS. Mle Urine Smples In totl, 172 mle urine specimens were evluted for performnce on the PANTHER. Dt re summrized in Tble 1. Forty-six urine smples were collected in the APTIMA trnsport medium nd initilly tested by the TIGRIS Am J Clin Pthol 2014;141: DOI: /AJCPFQ25SQVZAWHZ 399

4 Cheng nd Kirby / Evlution of the Hologic PANTHER System Tble 1 Comprison of Test Results for Mle Urine Specimens TIGRIS nd/or COBAS CT NG CT NG Chlmydi trchomtis (CT): 100% (95% score confidence intervl, 92.6% %) positive, 100% (96.8%-100.0%) negtive, nd 100% (97.7%-100.0%) overll greement. Neisseri gonorrhoee (NG): 100% (95% score confidence intervl, 92.1%-100.0%) positive, 100% (96.9%-100.0%) negtive, nd 100% (97.7%-100.0%) overll greement. (four CT+, one NG+, one CT+/NG+, nd 40 CT /NG ). On retesting by the PANTHER, 45 yielded identicl results. One discrepnt smple, designted s CT+/NG+ by n outside institution, tested CT /NG on the PANTHER nd on repet testing on both the PANTHER nd TIGRIS. This smple ws presumed to hve been misclssified by the outside lbortory nd ws therefore ressigned to the CT /NG ctegory for dt nlysis. In ddition, 126 mle urine smples initilly tested by the COBAS (47 CT+, 47 NG+, one CT+/NG+, nd 31 CT / NG ) were retested on both the PANTHER nd TIGRIS. Among COBAS NG+ smples, there were 13 discrepncies. All were initilly NG+ by the COBAS prior to frozen storge in the bsence of trnsport medium. However, on the PAN- THER, the results were s follows: nine NG+, three NG± (equivocl), nd one NG. Interestingly, further diminution of positivity ws observed on subsequent testing on the TIGRIS, with four NG± nd nine NG. To resolve these discrepncies, the smples were gin retested on the PANTHER, TIGRIS, COBAS, nd LC. On repet PANTHER testing, ll results were negtive, wheres on repet TIGRIS testing, ll but two were negtive. Interestingly, when retested on the COBAS, eight were NG+, four were NG±, nd one ws NG, very similr to initil PANTHER results. Lst, ll smples were positive when retested by LC ssys. However, four smples hd much delyed cycle threshold compred with LC when performed during the originl clinicl testing. The simplest explntion for these findings is tht urine smples were stored without preservtive per the COBAS testing protocol. Notbly, both the COBAS nd LC detect DNA in contrst to the AC2, which detects RNA. RNA is intrinsiclly much less stble thn DNA. Therefore, these discrepnt results likely reflect the preferentil degrdtion of RNA trget in smples tht do not hve gunidiniumbsed trnsport buffer to preserve RNA integrity. However, it is likely tht DNA ws lso degrded to some extent since even the DNA-bsed COBAS ssy showed some loss of positivity. By design, LC ssys, s described previously by our lbortory, 7 re more sensitive thn COBAS ssys since extrcted nucleic cids re concentrted by ultrfiltrtion prior to PCR rection setup (by fctor of 10-fold or more). This enhnced sensitivity, in combintion with DNA-bsed mplifiction ssy, provides resonble explntion for the bility of LC to detect NG in ll smples, even fter prtil degrdtion of trget. Since nine specimens greed on the PANTHER, initil testing on the COBAS, nd follow-up testing by LC, they were scored s concordnt NG+. The four remining PANTHER discrepncies (including one CT+/ NG+ smple) were excluded from subsequent nlysis since they were not relibly detected by the PANTHER or the TIGRIS comprtor. Similr issues confounded the nlysis of the COBAS CT+ mle urine smples. Among COBAS CT+ smples, there were five potentilly discrepnt smples. Among these, the PANTHER identified two s CT+ nd three s CT. The two CT+ smples were equivocl on the TIGRIS nd vribly equivocl, positive, or negtive on repet PAN- THER nd TIGRIS testing. Of the three PANTHER negtive smples, only one ws positive by the TIGRIS, but this smple becme negtive on TIGRIS repet nd positive on PANTHER repet. Tken together, two of the five discrepnt smples (initilly CT+ on COBAS) were resolved s concordnt CT+ since the initil results on the PANTHER greed with the originl COBAS results. As noted bove, the remining discrepncies cn best be explined by smple degrdtion in urine smples collected without preservtive nd vrible positivity due to stochstic effects. We therefore excluded the three PANTHER negtive, discrepnt specimens from subsequent nlysis. Among 31 CT /NG mle urine smples originlly tested by the COBAS, one smple ws negtive on the PAN- THER but positive on the TIGRIS. On subsequent retesting, the smple ws negtive on the TIGRIS nd borderline positive by the COBAS. This smple ws excluded from nlysis since CT ws not consistently detected by ny of the three methods. Femle Urine Smples In totl, 130 femle urine smples were evluted for performnce of the PANTHER. Dt re summrized in Tble 2. Ninety-seven smples were collected in the APTIMA urine trnsport tubes nd initilly tested on the TIGRIS t outside institutions (17 CT+, eight NG+, two CT+/NG+, nd 70 CT /NG ). Thirty-three were initilly tested by the COBAS (33 CT+). Among CT+ smples, there were six discrepncies. Three of these smples, originlly designted s CT+ on the TIGRIS by outside institutions, were confirmed negtive on repet testing by the TIGRIS nd PANTHER. These discrepncies likely were due to miscoding of the results on the smple tubes. Therefore, they were ressigned to the CT / 400 Am J Clin Pthol 2014;141: DOI: /AJCPFQ25SQVZAWHZ

5 AJCP / Originl Article NG ctegory for comprison. The remining three were COBAS CT+ smples tht tested CT+ on initil testing by the PANTHER. However, on testing by the TIGRIS, two of these were negtive nd one gve n invlid result. On retesting by the TIGRIS, the negtives remined negtive nd the invlid smple becme equivocl. The TIGRIS negtive smples were tested gin on the PANTHER nd COBAS. Both smples becme negtive on the PANTHER. On the COBAS, one becme CT nd the second remined CT+. As explined bove, these results re most likely explined by specimen degrdtion in unpreserved urine specimens. Tken together, results were resolved s concordnt CT+ since initil PANTHER CT+ results mtched the COBAS nd lter results could resonbly be explined by temporl trget degrdtion. Among NG+ smples, one ws negtive on the PAN- THER nd retesting on the TIGRIS. This smple ws ssumed to hve been miscoded by the outside lbortory nd ws resolved s CT /NG for specimen comprison. Notbly, the number of NG+ femle urine smples (ie, nine) ws much lower thn desired becuse of the low prevlence of NG in women in the greter Boston re. Therefore, dditionl NGspiked femle urine smples were used for dditionl evlution s discussed below. NG-Spiked Femle Urine Specimens Spiking studies re recommended method for evluting rre nlytes when diluted into the pproprite mtrix. 12 The nlyticl sensitivity clim for the PANTHER on pproved smple types is 50 CFU/mL. 14 Therefore, our gol ws to determine whether NG could be detected in distinct femle urine smples spiked with levels below this threshold. Hence, we spiked 48 COBAS CT clinicl femle urine smples with 12.5 CFU/mL of NG ATCC t concentrtion fourfold lower thn detection clims for pproved PANTHER specimen types nd the APTIMA positive control. Prior to spiking, smples were first tested by the PANTHER for both CT nd NG. Unexpectedly, one smple ws CT+ on the PANTHER (RLU = 412, low vlue, perhps ccounting for negtivity by the COBAS), positive result reconfirmed on the TIGRIS. After spiking, NG+ (RLU rnge, 339-1,248) ws detected in 47 smples. The remining smple ws equivocl for NG (RLU = 71) by the PANTHER nd borderline negtive when retested on the TIGRIS (RLU = 54). Since we hd purposefully used n NG+ concentrtion below the nlyticl sensitivity clim for the PANTHER, such result for single smple is not surprising. Specificlly, mong 48 spike smples, it is sttisticlly plusible tht some smples my hve stochsticlly received smller qunt of trgets just beyond the detection threshold of the ssy. This would ccount for equivocl/negtive results by both the PANTHER nd TIGRIS methods, whose RLU mesurements were essentilly the sme. Tble 2 Comprison of Femle Urine Specimens TIGRIS nd/or COBAS CT NG+ 9 0 CT 0 81 NG Chlmydi trchomtis (CT): 100% (95% confidence intervl, 92.7%-100.0%) positive, 100% (95.5%-100.0%) negtive, nd 100% (97.1%-100.0%) overll greement. Neisseri gonorrhoee (NG): 100% (95% confidence intervl, 70.1% %) positive, 100% (96.9%-100.0%) negtive, nd 100% (97.1%-100.0%) overll greement. Tble 3 Comprison of Combined Results for Urine Smples Including Spiked nd Unsexed Smples TIGRIS nd/or COBAS CT NG+ 109 b 0 CT b,c NG Chlmydi trchomtis (CT): 99.3% (95% confidence intervl, 96.0%-99.9%) positive, 100% (98.5%-100.0%) negtive, nd 99.7% (98.5%-100%) overll greement. Neisseri gonorrhoee (NG): 100% (95% confidence intervl, 96.6% %) positive, 100% (98.6%-100.0%) negtive, nd 100% (99.0% %) overll greement. b A single NG-spiked femle urine specimen with equivocl/borderline negtive reltive light unit vlues on the PANTHER nd TIGRIS ws excluded from nlysis. c Includes 46 NG-spiked femle urine smples tht were CT by the COBAS nd PANTHER but were not retested by the TIGRIS. Unsexed Urine Specimens Forty-seven deidentified urine smples, previously tested by the TIGRIS, were provided to us without designtion s to the sex of ptients (39 CT+ nd eight NG+). Among these smples, two designted CT+ t n outside institution tested CT on the PANTHER. The first smple ws CT on the TIGRIS nd PANTHER repet nd ws ressigned s CT / NG. The second smple on retesting ws positive on both the PANTHER (RLU = 115) nd TIGRIS (RLU = 102), respectively, very close to the positivity cutoff of 100. The very low level of positivity likely explined the vribility in trget detection. Nevertheless, the second smple ws tbulted s PANTHER flse-negtive bsed on initil testing results. The combined dt for femle, mle, unsexed, nd spiked urine specimens re shown in Tble 3. Femle nd Mle Swbs nd ThinPrep PreservCyt A more limited nlysis ws performed for pproved smple types, including 52 endocervicl swbs (33 CT+, eight NG+, one CT+/NG+, nd ten CT /NG ), two mle swbs (one NG+ nd one CT+/NG+), nd 15 ThinPrep PreservCyt smples (five CT+ nd ten CT /NG ), ll previously tested on the TIGRIS, with combined results shown in Tble 4. Notbly, the PANTHER filed to detect NG in one of nine Am J Clin Pthol 2014;141: DOI: /AJCPFQ25SQVZAWHZ 401

6 Cheng nd Kirby / Evlution of the Hologic PANTHER System Tble 4 Comprison of Combined Test Results for Femle Swbs, Mle Swbs, nd ThinPrep PreservCyt Specimens TIGRIS CT NG CT 0 29 NG 1 55 Chlmydi trchomtis (CT): 100% (95% confidence intervl, 90.6%-100.0%) positive, 100% (88.3%-100.0%) negtive, nd 100% (94.5%-100%) overll greement. Neisseri gonorrhoee (NG): 90.9% (95% confidence intervl, 62.8%- 98.4%) positive, 100% (93.5%-100%) negtive, nd 98.5% (91.9%- 99.7%) overll greement. Tble 5 Anlyticl Specificity in the Urine Mtrix Strin nd ATCC No. RLU CT NG Neisseri sicc Neg Neg Neisseri lctmic Neg Neg Neisseri mucos Neg Neg Neisseri cinere Neg Neg Stphylococcus ureus Neg Neg Stphylococcus sprophyticus Neg Neg Stphylococcus epidermidis Neg Neg Proteus vulgris Neg Neg Enterococcus fecium 3 Neg Neg Enterobcter cloce 5 Neg Neg Enterococcus feclis Neg Neg Pseudomons eruginos Neg Neg Lctobcillus 6 Neg Neg Escherichi coli Neg Neg Klebsiell pneumonie 6 Neg Neg Morxell ctrrhlis Neg Neg Streptococcus viridns 3 Neg Neg Streptococcus pyogenes Neg Neg Streptococcus glctie Neg Neg Cndid lbicns Neg Neg Citrobcter freundii 7 Neg Neg Acinetobcter bumnii Neg Neg Serrti mrcescens 8 Neg Neg Neisseri gonorrhoee Neg Pos ATCC, Americn Type Culture Collection; CT, Chlmydi trchomtis; Neg, negtive; NG, Neisseri gonorrhoee; Pos, positive; RLU, reltive light unit output from the APTIMA Combo 2 dul kinetic ssy. Clinicl isoltes. Tble 6 Reltive Limit of Detection Assessed by Replicte Testing of AmpliProbe CT/NG Control Serilly Diluted in Urine Mtrix % CT+ % NG+ Log Dilution TIGRIS PANTHER TIGRIS PANTHER 7.50E E E E E E E CT, Chlmydi trchomtis; NG, Neisseri gonorrhoee. NG+ endocervicl swb smples, both initilly (RLU = 8) nd on repet testing (RLU = 20). However, the smple ws positive when retested on the TIGRIS (RLU = 1,351) nd ws therefore resolved s PANTHER flse-negtive. Three femle swbs yielded invlid results on the PANTHER nd were excluded from nlysis since there ws insufficient smple for retesting. Only two mle swbs were vilble for testing: one ws CT /NG+ nd the other ws CT+/NG+. No NG+ ThinPrep PreservCyt specimens were vilble. All mle swb nd ThinPrep PreservCyt results were confirmed on the PANTHER. Anlyticl Specificity To exmine potentil effects of urine mtrix on nlyticl specificity, we seeded CT /NG urine with orgnisms frequently found s either urine contminnts or pthogens. As shown in Tble 5, no cross-rectivity ws observed when tested by the PANTHER, nd ll RLU were within the rnge found in CT /NG urine smples. Anlyticl Sensitivity in Urine Mtrix The reltive LODs for the PANTHER nd TIGRIS were determined through testing dilutions of the AmpliPROBE CT/NG control mteril in urine mtrix Tble 6. For CT, the PANTHER LOD, 4.9 (95% CI, ) log dilution, ws nominlly higher thn the TIGRIS LOD, 5.0 (95% CI, ) log dilution, s determined by probit nlysis. For NG, the LODs for the PANTHER nd TIGRIS were equivlent (5.9 log dilution). Overll, the nlyticl sensitivity ppered to be equivlent in urine mtrix within the power of the nlysis. Discussion We expected tht the PANTHER nd TIGRIS would hve lmost identicl performnce chrcteristics bsed on their use of the sme AC2 regents nd mplifiction conditions. However, it is conceivble tht the PANTHER utomtion might somehow compromise performnce reltive to other AC2-bsed ssys. Therefore, we evluted the performnce of the PANTHER compred with the well-estblished TIGRIS nd/or COBAS ssys. Tking into ccount evidence for specimen degrdtion in unpreserved urine nd likely misclssifiction of severl specimens t outside institutions, we found tht the PANTHER demonstrted nerly 100% positive nd negtive greement with the TIGRIS nd/or COBAS (Tbles 1-3) for urine smples. Furthermore, the PANTHER nd TIGRIS demonstrted n equivlent LOD within the power of nlysis. A more limited evlution ws performed for pproved specimen types in which greement between the PANTHER nd TIGRIS ws 100% for CT, CT+, nd 402 Am J Clin Pthol 2014;141: DOI: /AJCPFQ25SQVZAWHZ

7 AJCP / Originl Article NG nd 90.9% for NG smples with lrger 95% confidence intervls implicit in the smll numbers of smples tested (see Tble 4). During our study, we noted tht severl urine smples stored in the bsence of APTIMA trnsport medium (ie, without gunidinium-contining denturnt) showed loss of CT/NG detection, consistent with temporl degrdtion of the RNA trget during storge. We previously observed similr lbility in cervicl swb smples on storge, even for the more stble COBAS DNA trgets. 7 These observtions underscore the importnce of dding smples to trnsport medium s soon s prcticble to void specimen degrdtion nd flsenegtive results. Although we did not directly ssess clinicl sensitivity nd specificity of the PANTHER, we infer tht clinicl sensitivity nd specificity should be mintined reltive to other AC2-bsed ssys bsed on comprisons with the TIGRIS. Furthermore, the instrument utomtion nd user interfce were fcile. No dditionl opertor intervention is required fter loding regents nd smples. Moreover, multiple ssys cn be run t once, even on the sme specimen, providing intriguing possibilities for lbortory utomtion s dditionl ssys re pproved for the PANTHER. Tken together, the PANTHER provides n utomted, high-throughput ddition to methods for CT/NG detection with excellent performnce chrcteristics on wide rnge of smple types, including mle nd femle urine smples. Address reprint requests to Dr Kirby: Beth Isrel Deconess Medicl Center, 330 Brookline Ave, YA309, Boston, MA 02215; e-mil: jekirby@bidmchrvrd.edu. Acknowledgments: We thnk the clinicl microbiology lbortories t Brighm nd Women s Hospitl, Cmbridge Hospitl, Boston Medicl Center, Msschusetts Generl Hospitl, nd Needhm Hrvrd Vngurd for providing deidentified smples nd Hologic Gen-Probe for providing regents for this study. References 1. Dtt SD, Torrone E, Kruszon-Morn D, et l. Chlmydi trchomtis trends in the United Sttes mong persons 14 to 39 yers of ge, Sex Trnsm Dis. 2012;39: Stterwhite CL, Grier L, Ptzer R, et l. Chlmydi positivity trends mong women ttending fmily plnning clinics: United Sttes, Sex Trnsm Dis. 2011;38: Tuite AR, Jyrmn GC, Allen VG, et l. Estimtion of the burden of disese nd costs of genitl Chlmydi trchomtis infection in Cnd. Sex Trnsm Dis. 2012;39: US Preventive Services Tsk Force. Screening for chlmydil infection: U.S. Preventive Services Tsk Force recommendtion sttement. Ann Intern Med. 2007;147: Workowski KA, Bermn SM. Sexully trnsmitted diseses tretment guidelines, MMWR Recomm Rep. 2010;59: Americn College of Obstetricins nd Gynecologists Committee on Gynecologic Prctice. ACOG Committee Opinion No. 483: primry nd preventive cre: periodic ssessments. Obstet Gynecol. 2011;117: Cheng A, Qin Q, Kirby JE. Evlution of the Abbott RelTime CT/NG ssy in comprison to the Roche Cobs Amplicor CT/NG ssy. J Clin Microbiol. 2011;49: Golprin D, Tbrizi SN, Unemo M. Anlyticl specificity nd sensitivity of the APTIMA Combo 2 nd APTIMA GC ssys for detection of commensl Neisseri species nd Neisseri gonorrhoee on the Gen-Probe Pnther instrument. Sex Trnsm Dis. 2013;40: Gydos CA, Quinn TC, Willis D, et l. Performnce of the APTIMA Combo 2 ssy for detection of Chlmydi trchomtis nd Neisseri gonorrhoee in femle urine nd endocervicl swb specimens. J Clin Microbiol. 2003;41: Levett PN, Brndt K, Olenius K, et l. Evlution of three utomted nucleic cid mplifiction systems for detection of Chlmydi trchomtis nd Neisseri gonorrhoee in first-void urine specimens. J Clin Microbiol. 2008;46: Miyd CG, Born TL. A DNA sequence for the discrimintion of Neisseri gonorrhoee from other Neisseri species. Mol Cell Probes. 1991;5: Clrk RB, Lewinski MA, Loeffelhoz MJ, et l. Cumitech 31: Verifiction nd Vlidtion of Procedures in the Clinicl Microbiology Lbortory. Wshington, DC: ASM Press; AmpliPROBE CT/GC [pckge insert]. Sn Rmon, CA: Bio-Rd Lbortories; Gen-Probe APTIMA COMBO 2 Assy (PANTHER System), Rev. A [pckge insert]. Sn Diego, CA: Gen-Probe; Gen-Probe APTIMA COMBO 2 Assy Rev. A [pckge insert]. Sn Diego, CA: Gen-Probe; Clinicl nd Lbortory Stndrds Institute (CLSI). User Protocol for Evlution of Qulittive Test Performnce: Approved Guideline. 2nd ed. Wyne, PA: CLSI; Burd EM. Vlidtion of lbortory-developed moleculr ssys for infectious diseses. Clin Microbiol Rev. 2010;23: Am J Clin Pthol 2014;141: DOI: /AJCPFQ25SQVZAWHZ 403