PSA (total) EQA results: why they fail and what to check.
What is known to go wrong with PSA (total) in external quality assurance, how each problem shows itself in the results, and the checks to make, with every source listed. This is the PSA (total) monograph EQAWarden ships with. First check: Confirm which Atellica total PSA assay (PSA or tPSAII) is in use and compare with that Atellica group.
The limits, units and methods behind every PSA (total) result.
| RCPAQAP APS, Endocrine sub-modules | ±0.4 µg/L up to 5 µg/L, ±8% above. The value is verified; the programme it applies to (PSA sub-module or Tumour Markers) is inferred from the menus. |
|---|---|
| Units | µg/L. 1 ng/mL = 1 µg/L |
| Methods | Two-site sandwich chemiluminescent immunoassay, equimolar for free PSA and PSA-ACT: Atellica IM PSA (goat polyclonal acridinium label, mouse monoclonal covalently coupled to particles, water wash; WHO 96/670) and Atellica IM tPSAII (three mouse monoclonals including an unlabelled free-PSA antibody for equimolarity, biotinylated capture on streptavidin particles; WHO 17/100) (confidence: High)[1, 3] Other total PSA immunoassays calibrated to WHO 96/670 or to the manufacturer's traditional (Hybritech) calibration, about 25% higher (confidence: High)[5, 4] |
| Standardisation | JCTLM lists a purified PSA reference material (BCR-613, EU-JRC) but no reference measurement procedure was found[10]. Most methods are calibrated to WHO 96/670 (the 2nd IS 17/100 now exists), yet between-method differences of about -13% to +15% have been shown and Hybritech-calibrated results are about 25% higher than WHO-calibrated ones[4, 5, 9]. In a six-provider EQA study including RCPAQAP, Abbott, Beckman, Roche and Siemens Atellica met the desirable bias criterion for PSA on patient-pool samples[6]. Atellica PSA is traceable to WHO 96/670 and tPSAII to WHO 17/100[1, 3]. |
| Expected error mode | Proportional (calibration, lot, version); constant component at low PSA. |
| Biological variation | Within-subject CV 6.8% and between-subject CV 42.0%, from the EFLM database (meta-analysis, 1 study). |
| Desirable performance | Imprecision 3.4%, bias 10.64%, total allowable error 16.25%, derived from biological variation (Fraser). |
Platform details are for the Siemens Atellica, the platform EQAWarden's first release covers. Other platforms differ: check your own instructions for use.
11 ways a PSA (total) result goes wrong in EQA, and how to tell them apart.
Method-group differences despite WHO calibration
Why it happens. PSA methods differ through calibration (WHO versus Hybritech), antibody epitopes and response to free versus complexed PSA; Access, AxSYM, Centaur and Immulite ranged from 87% to 115% of each other[4]. Current major platforms, including Atellica, were within the desirable bias criterion on patient-pool EQA samples[6]. Atellica PSA gives y = 0.93 x ADVIA Centaur PSA + 0.238 ng/mL (n 105), reading higher than Centaur below about 3.4 µg/L and lower above[1].
Direction. method-specific
How it shows. method-group, persistent, proportional
Check. Compare with the Siemens Atellica group for the PSA assay in use; check whether a Hybritech-calibrated group is part of the all-method mean.
PSA and tPSAII are different assays with different WHO standards
Why it happens. First-generation PSA (CAL Q, WHO 96/670) and tPSAII (kit calibrators, WHO 17/100) differ in antibodies, standard, calibration intervals and measuring interval (0.01-100 versus 0.008-50 µg/L). A laboratory changing between them, running one on each analyser, or registered under the wrong method code should expect a step.
Direction. either
How it shows. step, method-group, proportional; begins when the assay changed
Check. Confirm which total PSA test definition each analyser runs (confirm in the laboratory's Atellica documentation) and that the RCPAQAP method code matches; look for a step at the change date.
Seminal-fluid PSA in control or EQA material
Why it happens. Both Atellica IFUs warn that material made by adding seminal-fluid PSA to serum can give assay-specific values because methods respond differently to free and complexed PSA, so assay-specific targets are needed. A material rich in free PSA spreads the method groups on that sample.
Direction. method-specific
How it shows. sample-specific, method-group
Check. Compare the spread between method groups for this sample with other samples; ask RCPAQAP about the PSA source if one sample is discordant.
High-dose hook effect is reached earlier on Atellica PSA
Why it happens. The first-generation Atellica PSA reports above range without hook only up to 2,500 µg/L; a published case read 78.0 µg/L on a sample of 9,520 µg/L, and the Atellica hook threshold was the lowest of four platforms compared (Cobas 17,000, VITROS 11,048, Alinity 48,000 µg/L)[12, 1]. tPSAII shows no hook to 25,000 µg/L[3]. EQA samples are normally far below these levels.
Direction. negative
How it shows. single, concentration-dependent; only a sample with a true value above the hook-free range
Check. If a very high EQA sample reads within range but lower than peers by a large factor, rerun at 1:10 and 1:100; check which PSA assay is in use.
New reagent lot and its master curve (PSA or tPSAII)
Why it happens. Each Atellica IM reagent lot carries a lot-specific master curve read from 2D barcodes; a new lot needs a new two-point calibration. Intervals in the IFU: lot calibration 29 (PSA) or 42 (tPSAII) days, pack calibration 28 (PSA) or 14 (tPSAII) days, onboard stability 28 (PSA) or 42 (tPSAII) days. The tPSAII pack interval (14 days) is much shorter than its onboard stability (42 days), so packs rely on a valid lot calibration; PSA needs a lot calibration every 29 days. A lot-to-lot difference that the two-point calibration does not correct shifts results on every analyser using that lot. The size of the shift can differ between patient samples and EQA or control material.
Direction. either (sign set by the lot difference)
How it shows. step, proportional, persistent; begins at the date the new reagent lot was first used and follows that lot
Check. Pull the reagent and calibrator lot history for the analysis date; compare IQC before and after the lot change and, if available, run the master curve material or a retained EQA sample on both lots.
Two-point recalibration or calibrator handling (PSA or tPSAII, CAL Q or tPSAII kit calibrators)
Why it happens. Atellica IM two-point calibration adjusts the lot master curve with a low and a high calibrator. CAL Q is lyophilised (human PSA in goat serum), reconstituted with 2.0 mL special reagent water; 21 days at 2-8 C, 8 hours at room temperature, no frozen claim. tPSAII calibrators are liquid (seminal-fluid PSA in buffer), 30 days at 2-8 C once opened and 8 hours on the system, and are bound to the kit lot. A reconstitution volume error, an evaporated or out-of-date opened calibrator, or a calibrator from the wrong lot shifts every result after that calibration on that analyser.
Direction. either
How it shows. step, proportional, instrument-specific; starts at a dated calibration on one analyser
Check. Open the calibration record for the run (date, calibrator lot, reconstitution or opening date and time, calibration acceptance); compare IQC before and after; compare with a second analyser calibrated separately.
Low PSA samples (post-prostatectomy range)
Why it happens. Within-lab CV of PSA was 8.5% at 0.06 µg/L and 3.2% at 0.32 µg/L; tPSAII 4.2% at 0.026 µg/L and up to 6.6% at 0.2 µg/L. The PSA-to-Centaur intercept of +0.238 ng/mL means a constant positive difference from Centaur-derived values at low PSA. The biochemical recurrence threshold of 0.2 µg/L sits in this region.
Direction. either: positive for the Centaur intercept; either for imprecision
How it shows. constant, concentration-dependent, method-group
Check. Judge a low sample against the APS absolute limit; regress the cycle's results on targets to separate constant from proportional error.
Chemiluminescent signal or wash fault in a sandwich assay
Why it happens. The assay is a two-site sandwich with a direct relationship between analyte and relative light units (RLU). A loss of signal (trigger reagents, luminometer, particle loss at washing) reads low, while poor washing of unbound label reads high; competitive assays on the same analyser move the opposite way. The assay chart lists a water wash for PSA (as for Fer) and the APW1 probe wash, which lasts only 14 days on board; tPSAII uses Atellica IM Wash, so PSA and ferritin may move together apart from the IM Wash assays (confidence: Medium, inferred from the chart).
Direction. either: negative for a loss of signal; positive for residual unbound label
How it shows. instrument-specific, survey-wide, analyte-group:im_sandwich; sandwich assays move one way and competitive assays the other on the same analyser
Check. Check whether other Atellica IM sandwich assays on the same analyser moved in the same direction and competitive assays in the opposite direction; review trigger reagent, wash fluid and probe wash onboard times, luminometer checks and maintenance logs.
Onboard dilution
Why it happens. PSA dilutes onboard 1:2 to 1:100 with Multi-Diluent 2 (recovery 94-109%, mean 102%); tPSAII 1:5 to 1:500 (designed recovery 90-110%). A wrong manual dilution factor gives a large proportional error on that sample.
Direction. either
How it shows. single, concentration-dependent, proportional
Check. For a high sample, check the dilution flag and factor and Multi-Diluent 2 onboard age (28 days).
Test definition update left operator-defined parameters unchanged
Why it happens. Siemens issues new Atellica IM test definition versions (PSA test definition 1.3 added alternate units with a default conversion factor (ng/mL to µg/L is 1:1)). Operator-defined parameters are not updated automatically when a test definition is installed, so a dilution point, unit or conversion factor set by the laboratory can differ from the new default.
Direction. either
How it shows. step, instrument-specific; starts at the software or test definition update
Check. Compare the installed test definition version, reporting unit and conversion factor on each analyser with the current assay chart and the enhancement notice; check the date of the last test definition installation.
Heterophile or human anti-animal antibodies
Why it happens. The IFU says the assay is designed to minimise heterophilic antibody interference but that such antibodies can still give falsely high or low results. tPSAII was tested with HAMA 2640 µg/L and rheumatoid factor 1500 IU/mL with 10% or less change. These antibodies belong to individual donors; a pooled EQA material is unlikely to carry them at an effective level unless it comes from a single donor.
Direction. either
How it shows. single, sample-specific; one sample reads off in one method group and not in others
Check. If a single-donor EQA sample is off in the Atellica group only, ask RCPAQAP about the material source; retest after a heterophile blocking step or on another platform if the laboratory keeps the sample.
The checks, in order.
- Confirm which Atellica total PSA assay (PSA or tPSAII) is in use and compare with that Atellica group.
- Check calibration and lot dates (PSA lot 29, pack 28 days; tPSAII lot 42, pack 14 days) and calibrator age.
- For a very high sample reading implausibly low, rerun diluted (Atellica PSA hook from about 2,500 µg/L).
- For a low sample, judge the absolute difference against the APS absolute limit.
- If PSA and ferritin (water-wash assays) or several IM sandwich assays moved together on one analyser, review wash, probe wash and signal records.
- Units: µg/L and ng/mL are numerically identical.
Unit traps
- µg/L and ng/mL are numerically identical; a ratio of 1000 suggests ng/L.
EQAWarden runs these checks as rules on every survey, ranks the likely causes with their evidence, and the reviewer decides. All analytes · The 58 rules · The whole knowledge base · How an EQA investigation runs
Questions about PSA (total) EQA.
What is the RCPAQAP allowable performance specification for PSA (total)?
In the Endocrine sub-modules (ACE; AMH; Bone markers P1NP/beta-CTX; PSA total/free; Red cell folate) programme it is ±0.4 µg/L up to 5 µg/L, ±8% above, evaluated at the target. RCPAQAP can change its specifications, so confirm the current table at each enrolment year.
What is a common cause of a failed PSA (total) EQA result?
Method-group differences despite WHO calibration. PSA methods differ through calibration (WHO versus Hybritech), antibody epitopes and response to free versus complexed PSA; Access, AxSYM, Centaur and Immulite ranged from 87% to 115% of each other. Current major platforms, including Atellica, were within the desirable bias criterion on patient-pool EQA samples. Atellica PSA gives y = 0.93 x ADVIA Centaur PSA + 0.238 ng/mL (n 105), reading higher than Centaur below about 3.4 µg/L and lower above.
What should be checked first when a PSA (total) EQA result fails?
Confirm which Atellica total PSA assay (PSA or tPSAII) is in use and compare with that Atellica group.
How are PSA (total) units converted?
1 ng/mL = 1 µg/L.
Where every fact on this page came from.
- Siemens Healthineers. Atellica IM Prostate-Specific Antigen (PSA) [OUS], Instructions for Use, part 10995416 Rev. 07, 2025-06. Siemens Document Library id 1137070.
- Siemens Healthineers. Atellica IM and Atellica CI Calibrator Q (CAL Q), Instructions for Use, part 11204925 Rev. 03, 2023-05. Siemens Document Library id 1003221.
- Siemens Healthineers. Atellica IM total PSA II (tPSAII) [OUS], Instructions for Use, part 11202945 Rev. 03, 2026-07. Siemens Document Library id 1187137.
- Stephan C, Klaas M, Muller C, Schnorr D, Loening SA, Jung K. Interchangeability of measurements of total and free prostate-specific antigen in serum with 5 frequently used assay combinations: an update. Clin Chem 2006;52:59-64. With Access tPSA as 100%, other assays ranged from 87% (AxSYM, ADVIA Centaur) to 115% (Immulite).
- Stephan C, Kahrs AM, Klotzek S, et al. Toward metrological traceability in the determination of prostate-specific antigen (PSA): calibrating Beckman Coulter Hybritech Access PSA assays to WHO standards compared with the traditional Hybritech standards. Clin Chem Lab Med 2008;46:623-629. WHO 96/670 calibration gave tPSA about 25% lower than Hybritech calibration.
- van Rossum HH, Holdenrieder S, Ballieux BEPB, Badrick TC, et al. Investigating the current harmonization status of tumor markers using global external quality assessment programs: a feasibility study. Clin Chem 2024;70:669-679. Six EQA providers including RCPAQAP, 2020-2021; for PSA the mean differences of Abbott Alinity, Beckman DxI, Roche Cobas and Siemens Atellica from the consensus mean on patient-pool EQA samples were within the desirable BV bias criterion.
- Filella X. Standardization of prostate-specific antigen assays: impact on reference intervals and clinical decision thresholds. Tumour Biol 2026;48:14230380261466593 (review: clinically relevant between-platform differences persist despite WHO 96/670; within-subject variation 6-13%). Author declares an agreement with Siemens.
- Filella X, Gonzalez-Escribano C, Rodriguez-Garcia M, Medina-Esteban N, Fernandez-Galan E. Analytical verification and comparative assessment of the new Atellica IM high-sensitivity prostate specific antigen assay. Clin Chem Lab Med 2026;64:913-921. LoD 0.01 µg/L, LoQ 0.028 µg/L; accurate up to 13,311 µg/L with minimal hook effect.
- DiaSorin. LIAISON PSA II Gen product page: standardised against the WHO 2nd International Standard for prostate specific antigen (human), NIBSC code 17/100, 2019. Manufacturer statement used only to identify WHO 17/100 as the 2nd IS for PSA.
- JCTLM Database of higher-order reference materials and reference measurement procedures. Keyword searches (troponin, troponin I, cardiac troponin, ferritin, prostate, PSA, natriuretic, BNP) through the public API https://www.jctlmdb.org/backend/api/jctlm/searchRm and searchRmp (multipart fields page_number, nb_results, keywords), 2026-09-29. Found only C1RM_P4, BCR-613 prostate specific antigen (purified PSA, EU-JRC, certified mass 70.8 ug per unit); none found for troponin I or T, ferritin, BNP or NT-proBNP. Absence in a keyword search is reported as 'none found', not as proof of non-listing.
- Siemens Healthineers. Atellica Solution IM Assay Chart, part 11314389 Rev. 05, 2025-12. Siemens Document Library id 1151608. https://content.doclib.siemens-healthineers.com/rest/v1/view?document-id=1151608
- Benamour M, Brouwers P, Nevraumont A, Roy T, Bayart JL. Double trouble: unmasking two hook effects on Siemens Atellica - total PSA and total hCG assays. Pract Lab Med 2024;39:e00366. Atellica IM1600 total PSA read 78.0 ng/mL on a sample of 9520 ng/mL after 1:100 dilution; insert hook thresholds compared: Atellica 2500, Cobas 17,000, Alinity 48,000, VITROS 11,048 ng/mL.
- Jassam N, Jones CM, Briscoe T, Horner JH. The hook effect: a need for constant vigilance. Ann Clin Biochem 2006;43:314-317. doi:10.1258/000456306777695726
- Siemens Healthineers. Atellica Solution Operator's Guide, software version 1.31, part 11069101 Rev. 13, 2025-10.
- Miller WG, Erek A, Cunningham TD, Oladipo O, Scott MG, Johnson RE. Commutability limitations influence quality control results with different reagent lots. Clin Chem 2011;57:76-83. doi:10.1373/clinchem.2010.148106
- Wild D (ed.). The Immunoassay Handbook: Theory and Applications of Enzyme Immunoassay, Electrochemiluminescence and Related Techniques. 4th ed. Oxford: Elsevier; 2013. Cited as textbook knowledge for competitive versus sandwich signal-response relationships; not re-read in this review.
- Siemens Healthineers. Atellica IM Analyzer Test Definition Enhancements, Customer Information, part 11314997 Rev. 17, 2026-05. Siemens Document Library id 1173692. https://content.doclib.siemens-healthineers.com/rest/v1/view?document-id=1173692
- Kricka LJ. Human anti-animal antibody interferences in immunological assays. Clin Chem 1999;45:942-956. doi:10.1093/clinchem/45.7.942
- Tate J, Ward G. Interferences in immunoassay. Clin Biochem Rev 2004;25:105-120 (AACB journal). https://pmc.ncbi.nlm.nih.gov/articles/PMC1904417/
- EFLM Biological Variation Database (Aarsand AK et al., EFLM WG-BV). https://biologicalvariation.eu/ - meta-analysis values read from https://biologicalvariation.eu/api/meta_calculations and individual studies from https://biologicalvariation.eu/api/bv_specifications, retrieved 2026-09-28. https://biologicalvariation.eu/
- Fraser CG. Biological Variation: From Principles to Practice. Washington DC: AACC Press; 2001 (textbook; formulae desirable CVa<=0.5CVi, bias<=0.25*sqrt(CVi^2+CVg^2), TEa=1.65*CVa+bias).
- RCPAQAP. Chemical Pathology Analytical Performance Specifications (web page; dateModified 2025-03-05). https://rcpaqap.com.au/resources/chemical-pathology-analytical-performance-specifications/
See EQAWarden review a PSA (total) result.
Twenty minutes on a video call, on a fictional demo year: the score, the findings, the ranked causes and the signed review.