HPLC Carryover: Testing, Causes and Prevention

A peak in a blank is a finding to investigate, not proof that the needle wash has failed. HPLC carryover is residual response associated with material introduced during an earlier analysis; contamination may instead enter with the blank, vial, solvent, mobile phase or environment. Establish the sequence history, challenge the suspected mechanism with appropriate controls, and judge the response against the analytical decision it could affect. A small percentage of a high standard can still be important for a low-level impurity. The useful outcome is a documented explanation, an effective control and an assessment of affected results—not merely a clean chromatogram after repeated washing.

Carryover and contamination are different hypotheses

Start with the original chromatograms, injection order, sample concentrations, vial positions, method versions and instrument events. Record whether the suspect peak occurs at the analyte retention time and whether its response varies after high samples. Retention-time agreement alone does not establish identity: another constituent can coelute. Where appropriate, compare spectral or other available identification evidence without assuming that every detector can distinguish the substances.

Carryover often has a relationship with previous loading and may decline in successive blanks. A roughly constant response in blanks prepared independently suggests a continuing source, but neither pattern is conclusive. Persistent adsorption can survive a previous sequence, and contaminated solvent can produce changing responses during a gradient. Define what “blank” means: injection solvent, mobile phase, a matrix blank or a preparation blank answer different questions. An unused vial is not automatically a contamination-free control.

Design a diagnostic sequence around a question

Prepare a written test plan before changing wash conditions. Include the initial system state, a relevant pre-challenge blank, a justified high sample or standard, and post-challenge blanks suited to the hypothesis. Add independent blank preparations, relevant matrix controls or low-level samples when they help assess interference. Choose the high challenge from the method's range and realistic sample exposure; an extreme solvent or concentration outside that range can create an unrelated failure mechanism.

Specify the number and arrangement of injections from the objective, expected variability and persistence, rather than treating one standard sequence as universal. Repeated blanks can describe decay; they do not by themselves identify the contaminated component. A method cycle without an injection, if available as a supported function, can help distinguish contributions, but it may change valve operation and the fluid path. Retain the complete diagnostic sequence and explain those differences.

Quantify carryover against the analytical decision

Define the measured response, integration rules, reporting reference and treatment of background in advance. Area ratios can describe response, but converting them into concentration requires an appropriate response relationship. Detector saturation, different injection volumes or matrix effects can invalidate a simple ratio. Do not subtract a variable blank response merely to obtain a passing result; any correction needs a scientific basis and an authorised procedure.

Illustrative calculation—not experimental data: a high standard gives 1,000,000 area units, the following blank gives 100, and a relevant low-level standard gives 1,000. The blank is 100/1,000,000 × 100 = 0.01% of the high response but 100/1,000 × 100 = 10% of the low response. These are independently checked ratios, not acceptance limits. They show why the denominator, low-end performance and possible decision error matter more than an isolated percentage.

Set the criterion in relation to the intended use, reportable range, impurity or analyte levels, selectivity and acceptable interference. ICH Q2(R2) supplies a validation framework; Q14 places controls within the analytical procedure. ICH M10 discusses carryover in its specified bioanalytical context, including biological study samples. Its criteria must not be imported automatically into every pharmaceutical QC HPLC method. The Clouser-Roche study also illustrates the importance of considering effects on subsequent sample results rather than a detached pass/fail number.

Investigate the autosampler without guessing a wash recipe

Map the sample-contacting surfaces for the actual configuration: needle, seat, loop, valve passages, connecting lines and wash arrangements. Internal and external needle washing do not necessarily reach the same surfaces. Check wash-solvent identity, preparation and expiry, liquid availability, approved settings, waste handling and visible abnormalities. Compare the method with the configuration actually used, including any recent replacement or software-controlled change.

Choose a wash strategy from analyte solubility and adsorption, solvent miscibility, material compatibility and the manufacturer's model-specific instructions. A stronger organic solvent is not invariably better: precipitation, seal incompatibility or poor removal of an ionic residue can make the problem worse. Change one justified factor at a time when possible. The Vallano case study demonstrates that the injection mechanism can matter; it does not establish a universal remedy or authorise a different injection mode for every method.

Separate an approved operational setting from a change affecting instrument performance or method validity. A new loop, valve configuration or injection mode may require impact assessment and targeted verification or qualification. Before access to fluidic parts, follow the approved procedure for pressure, solvents, electrical energy and hot components. Do not loosen pressurised fittings, bypass interlocks or perform service-only work as a diagnostic shortcut.

Distinguish column memory from other fluid-path surfaces

Adsorption and delayed desorption can occur on stationary phases and other wetted materials. Difficult compounds may produce persistent or concentration-dependent behaviour, and more than one surface can contribute. The Yamagaki and Yamazaki study on a particular biomolecule illustrates that investigating only one component can miss additional contributions. Its observations cannot be converted into a cleaning recipe for unrelated analytes, columns or instruments.

Compare history, loading, solvent compatibility and the timing of release. If an authorised procedure permits testing with a different column or a bypass arrangement, recognise that retention, dilution, backpressure and detector exposure change. A missing peak after removal of the column does not prove that the column was the sole source: the signal may no longer be separated or recognisable. Preserve baseline conditions and use controls that make the comparison interpretable. Do not reverse a column or apply aggressive cleaning unless that specific configuration permits it.

Demonstrate correction under relevant conditions

Define success before implementing the selected action. Repeat a justified challenge with the same response reference and relevant low-level controls, considering the highest credible exposure, representative matrices, sequence position and normal waiting periods. If an extra wash or blank is proposed as a control, demonstrate how it protects the next reportable sample and what happens when the sequence pauses or resumes. A single clean blank immediately after extensive cleaning is weak evidence for sustained routine performance.

Simulated case A: the initial blank is clean; a peak appears after a high standard and declines in later blanks. Investigate a load-related memory mechanism, then test the selected correction with the same challenge. This pattern supports a direction of investigation, not an automatic needle diagnosis. Simulated case B: the peak is already present before the new high standard. Investigate independently prepared blanks, mobile phase, containers and prior instrument history first. Carryover from an earlier sequence remains possible; the new standard cannot explain an earlier observation.

Compare the corrected condition with the original problem and document residual uncertainty. If an intervention removes the signal but several variables changed together, do not claim a uniquely established root cause. An effective containment measure and a confirmed root cause are different conclusions. Monitor the relevant indicator after return to routine use and define who reviews an adverse trend.

Assess affected data and prevent recurrence

Reconstruct which samples followed relevant high exposures and whether the affected response overlaps a reportable analyte or impurity. Consider both magnitude and direction of possible bias, sequence placement, intervening controls and the sensitivity of the analytical decision. Extend the review to earlier work when the evidence supports a longer persistence period. Do not automatically invalidate an entire batch, release it because an SST passed, or declare a sample unaffected solely because its chromatogram looks acceptable.

Use the laboratory's investigation and data-integrity procedures to retain original injections, failed controls, reinjections, changes and the reasoning for any additional analysis. A reinjection is a planned investigative action, not a means of selecting a favourable result. Record who assessed sample impact and who approved the disposition. FDA's data-integrity guidance supports complete, reliable records; it does not supply a carryover limit for this method.

• Analyst: confirm blank identity, wash readiness and the approved sequence; escalate unexpected peaks without deleting data.
• Method owner: justify challenge levels, criteria and sequence controls; revisit them when analytes, matrices or reporting levels change.
• Instrument owner and service: maintain configuration and intervention records; establish the scope of post-intervention checks.
• Reviewer or quality unit, as locally assigned: assess the investigation, affected results and effectiveness monitoring before accepting closure.

This is a decision framework, not an authorised cleaning SOP. Its practical output should be a traceable set of hypotheses, discriminating tests, justified limits, corrective evidence and a sample-impact decision. Those elements make prevention transferable to routine work without pretending that every HPLC system has the same contamination mechanism.

Sources and scope

Research articles support mechanism-specific examples; their abstracts and accessible publisher material were consulted. FDA documents are final guidance in the stated US/ICH context, not universal operating procedures. The diagnostic matrix and simulated cases are original GuideGxP editorial tools.

1. Vallano PT et al. Elimination of autosampler carryover in a bioanalytical HPLC-MS/MS method: a case study. J Pharm Biomed Anal. 2005;36:1073–1078. DOI: 10.1016/j.jpba.2004.09.010.
2. Clouser-Roche A et al. Beyond pass/fail: a procedure for evaluating the effect of carryover in bioanalytical LC/MS/MS methods. J Pharm Biomed Anal. 2008;47:146–155. DOI: 10.1016/j.jpba.2007.12.019.
3. Yamagaki T, Yamazaki T. Troubleshooting Carry-Over in the LC-MS Analysis of Biomolecules: The Case of Neuropeptide Y. Mass Spectrom (Tokyo). 2020;8(2):S0083. DOI: 10.5702/massspectrometry.S0083.
4. FDA/ICH M10. Bioanalytical Method Validation and Study Sample Analysis. Final guidance, November 2022, §§1.3, 3.2.6.
5. FDA/ICH Q2(R2). Validation of Analytical Procedures. Final guidance, March 2024, §3.1.
6. FDA/ICH Q14. Analytical Procedure Development. Final guidance, March 2024, §6.
7. FDA. Data Integrity and Compliance With Drug CGMP: Questions and Answers. Final guidance, December 2018.

Related reading

• System suitability: criteria and failed tests
• Mobile-phase preparation and contamination controls
• Distinguishing baseline noise, drift and peaks