Lot change versus supplier change: two different risk profiles

A 6-point drift in the internal standard response factor showed up on a Tuesday afternoon run, four weeks after we switched to a new lot of 11 mm septa from the same supplier. Same part number, same certificate, same 45 Shore A silicone face. The system suitability check on the previous lot had been tight: RSD of 0.4% across six injections. On the new lot it was 1.8%, and the response factor had moved enough that the analyst flagged it before the batch was released. That single number is where this article starts, because the question that followed (“do we revalidate?”) is the one that consumes more compliance time than any other change-control decision in a QC lab, and the answer is not the same for a supplier change as it is for a lot change on the same drawing.

Revalidation triggers when an hplc septa supplier or lot changes are not written into most method validation reports, which is the root of the problem. USP <1225> tells you to control changes to the method, but it does not enumerate consumable lots. The change control entry has to do that work, and an inspector will read it against the method’s own criticality assessment. Get the trigger logic right and the entry writes itself. Get it wrong and you either over-validate (wasting six weeks of analyst time on a lot that only needed a verification run) or under-validate (releasing a method that has quietly shifted its baseline).

A lot change from the same supplier, same drawing, same material lot lineage, is a low-to-moderate risk event. The manufacturing parameters that matter for a septum (curing temperature, post-cure time, mold release agent, slit geometry if slit) are nominally identical. What drifts between lots is small: dimensional tolerance stack-up on the 11 mm OD, durometer variation of 2 to 4 Shore A, and extractables profile from a different silicone polymer batch. Those are real, and they are why a lot change is not a no-op, but they are bounded.

A supplier change is a different animal. New mold, new cure oven, new release agent, possibly a different base polymer grade, and almost certainly a different dimensional Cpk. I have seen a supplier change move the puncture count on an 11 mm silicone/PTFE septum from 180 injections to 95 on the same nominal part. That is not a drift, it is a different product. Treat it as a method change until proven otherwise.

The decision rule I use, and the one that survives audit, is this: a supplier change is a revalidation trigger by default; a lot change is a verification trigger by default, with an escalation path to revalidation if the verification fails. Write that rule into the method’s change-control section before the next supplier change happens, not after.

What counts as a supplier change

Any of the following is a supplier change, even if the part number and drawing are identical:

• Change of manufacturing site, including a move to a different building on the same campus.
• Change of base polymer grade or vendor, even at the same durometer.
• Change of mold or mold cavity count.
• Change of post-cure schedule (temperature, time, or atmosphere).
• Change of release agent or a switch to a release-free mold.
• Any change to the PTFE lamination process for laminated septa.

If the supplier cannot confirm in writing that none of these changed, treat the shipment as a supplier change. The HPLC septa you receive from a new source will pass a visual AQL and still fail a baseline drift check, because AQL plans do not measure extractables.

What counts as a lot change

A lot change is a new manufacturing lot from the same supplier, same site, same mold, same cure schedule, with a certificate of analysis that shows the two critical dimensions (OD and thickness) and the durometer within the historical Cpk band. If the CoA shows a durometer shift of more than 3 Shore A from the prior lot, escalate. If the CoA is missing the durometer, escalate. If the CoA is a generic “conforms to drawing” statement with no measured values, that is not a lot change, that is an unverified shipment, and you should treat it as a supplier change until you have measured data.

The verification package that replaces revalidation for a lot change

You do not need to re-run the full validation for a lot change. You need a verification package that answers one question: did the method’s performance characteristics stay inside the acceptance criteria already defined in the validation report? That package has four parts, and each part takes less than a shift to run.

Part 1: system suitability on the new lot

Run the method’s own system suitability test (SST) on the new lot, using the same standard, same column, same instrument, same analyst if possible. Compare against the SST limits in the method, and against the historical mean and standard deviation from the last 20 lots. The acceptance criterion is not “passes SST.” It is “passes SST and the response factor is within 2 standard deviations of the historical mean.” A pass that sits at 2.9 SD is a warning, not a pass.

Part 2: baseline noise and drift

Run a blank injection (mobile phase only) at the method’s detection wavelength for the full run time. Measure baseline noise (peak-to-peak over a 1-minute window) and baseline drift (slope over the run). Compare to the validation report’s baseline acceptance criteria. A lot that bleeds more than the prior lot will show up here as elevated noise at low wavelengths (210 to 220 nm is where silicone extractables and mold release agents show). If noise rises more than 30% over the prior lot, escalate to revalidation. This is the single most sensitive lot-change check I know of, and it takes 20 minutes.

Part 3: carryover and injection precision

Run six replicate injections of a mid-level standard and calculate %RSD. Then run a blank immediately after the highest standard and measure carryover at the analyte retention time. The septum contributes to carryover through two mechanisms: coring debris that lodges in the needle seat, and extractables that accumulate on the column head. Both show up as a rising baseline or a ghost peak. Acceptance: %RSD no worse than the validation report’s precision criterion (typically 1.0% for a well-behaved method), carryover no higher than the method’s LOQ.

Part 4: puncture count verification (destructive)

Take 10 septa from the new lot, puncture each 100 times with the method’s needle gauge at the method’s injection volume and tray configuration, and inspect for coring. Count visible cores and measure the puncture channel diameter under a 10x loupe. If more than 1 in 10 shows a core, or if the channel diameter exceeds the needle OD by more than 0.05 mm, the lot fails. This is the check that catches the supplier who changed their silicone cure and quietly halved the puncture life.

When a lot change escalates to full revalidation

Escalate to revalidation (accuracy, precision, linearity, LOD/LOQ, robustness, at minimum) if any of the following is true:

• Baseline noise rises more than 30% over the prior lot at the method wavelength.
• Response factor shifts more than 2 SD from the historical mean.
• Carryover exceeds the method LOQ.
• Puncture count drops below the method’s validated injection count.
• The lot’s CoA shows a durometer shift greater than 3 Shore A or a dimensional shift outside the historical Cpk band.

If none of those fire, the verification package is your change-control evidence and you release the lot. If one fires, you have a documented reason for revalidation that an inspector will accept without argument, because the trigger is a measured number, not a judgment call.

The change-control entry an inspector will accept

This is where most labs lose time. The verification work is done, the numbers are good, and the change-control entry says “new lot of septa, verified, no impact.” That entry will get a finding, because it does not show the reasoning. An inspector reading it cannot tell whether you assessed the risk or skipped the assessment. The entry needs six elements, in this order.

1. Description of the change, with the part number and lot

State the old lot, the new lot, the supplier, the part number, the drawing revision, and the CoA number. If the supplier changed, name the old and new supplier. If the site changed, name both sites. Vague entries (“new shipment of septa”) are the ones that get questioned.

2. Risk assessment against the method’s criticality

Reference the method’s own criticality assessment (every validated method has one, or should). State whether the septum is a critical consumable for this method: does it contact the sample, does it sit in the injection path, does the method’s detection wavelength sit below 230 nm where extractables matter? If yes, the risk is high and the verification package is mandatory. If no, the risk is low and a CoA review plus SST is sufficient. Write the reasoning, not just the conclusion.

3. The verification data, summarized with numbers

Put the actual numbers in the entry: SST response factor, %RSD, baseline noise in µAU, carryover in % of LOQ, puncture count. Attach the chromatograms. An entry that says “SST passed” without numbers is a claim, not evidence.

4. The acceptance criteria, stated explicitly

State what “pass” meant before you ran the test. “Response factor within 2 SD of historical mean (mean 1.02, SD 0.015, acceptance 0.99 to 1.05)” is an acceptance criterion. “Acceptable” is not. The analytical testing group should sign off on the criteria before the data is generated, not after.

5. The conclusion and the release decision

One sentence: “Verification passed all four acceptance criteria; lot released for use; no revalidation required.” Or: “Baseline noise rose 42% over prior lot; revalidation initiated per SOP-XXX.” Either is fine. Silence is not.

6. The approval signatures, dated

QC manager, method owner, and QA. Dated. The date matters because the inspector will check that the verification was completed before the lot was released, not after.

Two field cases where the trigger logic mattered

Case 1: the supplier change that looked like a lot change

A contract lab switched septum suppliers to save 18% on unit cost. The new supplier shipped a CoA that matched the drawing on OD and thickness, and the lab ran SST only. SST passed. Three weeks later, a client method at 214 nm started failing system suitability on baseline noise. The investigation traced it to mold release agent bleeding from the new septa. The measurement that identified it was a blank injection at 214 nm compared side-by-side with the old lot: noise was 2.4x higher. The fix was a full revalidation of the affected method (six client methods, actually) and a return to the original supplier. The lesson: a supplier change with a matching CoA is still a supplier change, and SST alone does not catch extractables.

Case 2: the lot change that should have escalated

A pharma QC lab received a new lot from their existing supplier. The CoA showed durometer at 48 Shore A against a historical mean of 44. The lab ran the verification package, and SST passed. But the puncture count check showed 110 injections before coring, against a validated 150. The lab escalated to revalidation, which confirmed the method’s precision was still acceptable but the injection count in the method’s sample prep section needed to be revised downward. The fix was a method revision (not a revalidation failure) and a supplier conversation about cure schedule. The lesson: the CoA durometer value is the earliest warning you get, and it costs nothing to read it.

Decision checklist: verification or revalidation

Operational precautions that keep the verification honest

Three things will invalidate a verification package faster than a bad lot will, and all three are procedural.

Do not change the column at the same time. If the verification run uses a new column, you cannot attribute a baseline shift to the septum. Same column, same lot, same install date.

Do not change the vial or cap. The vial closures and the HPLC vial contribute their own extractables and their own seal geometry. Hold them constant. If you must change them, run a separate verification.

Torque the cap to the method’s specified value. Most methods do not specify torque, which is a gap. For an 11 mm crimp cap, 1.5 to 2.0 N·m is the working range; for a screw cap, hand-tight plus a quarter turn. An over-torqued cap deforms the septum and changes the puncture channel geometry, which changes the coring behavior, which invalidates your puncture count check. Write the torque value into the method.

What to ask the supplier before the change happens

The verification package is cheaper than revalidation, but the cheapest option is a supplier who tells you about the change before you receive the shipment. Ask for four things in the supply agreement, and put them in writing:

• Advance notification of any change to site, mold, cure schedule, or base polymer, with a minimum of 90 days notice.
• A CoA with measured OD, thickness, and durometer for every lot, not a “conforms” statement.
• The Cpk on OD and thickness for the last 20 lots, on request.
• A change-control contact who will answer a technical question in writing within five business days.

Suppliers who can meet those four will also tell you when their silicone vendor changes, which is the change that actually moves your baseline. Suppliers who cannot meet them will ship you a lot that passes AQL and fails SST, and you will find out three weeks later on a client method. The Agilent and Thermo Fisher consumable lines publish change notifications because their customers require it; smaller suppliers often do not, which is a risk you are buying along with the discount. For reference methods and the regulatory frame around change control, EPA SW-846 and the technical notes at Restek are worth reading before you write the SOP, because they show how the analytical community treats consumable changes in practice.

The decision rule to write into your SOP

Treat every supplier change as a revalidation trigger and every lot change as a verification trigger, escalate on measured numbers (baseline noise up more than 30%, response factor outside 2 SD, carryover above LOQ, puncture count below the validated injection count, durometer shift above 3 Shore A), and write the change-control entry with the six elements in order so the reasoning is visible. The verification package costs one shift. The revalidation costs two to four weeks. The finding costs more than either, and it arrives eighteen months later when the lot is long gone and the analyst who ran the SST has left the company.

References & further reading

Related resources from our lab consumables network:

• Hplc Septa at hplcsepta.com
• Vial Closures at vialsclosures.com
• Analytical Testing at testinganalytical.com
• Hplc Vial at vialforhplc.com

External references and standards:

• Agilent chromatography consumables and columns
• Thermo Fisher chromatography and sample handling
• U.S. EPA SW-846 test methods for waste and materials
• Restek technical resources for chromatography