Additional Capabilities in Cylinder Plate 5+1 Bioassay

Bioassay — Cylinder-Plate 5+1 Assay

Capabilities covered: Independent Replicate Sets ("Replicate" column) and Nominal vs. Actual Reference Dose ("Nominal Reference Dose")

This tool estimates the potency of a test/unknown sample against a five-concentration standard curve, using the classical "corrected mean" method for cylinder-plate (zone of inhibition) microbial assays. It has two optional settings that extend how a single assay run's sample side can be handled; both are covered in this document. Unlike the error-term options documented for the other bioassay tools, these two settings were not added to reconcile a discrepancy against a specific published example — they are functional extensions of the base method, checked here against USP <81> guidance and against internal self-consistency, rather than against a single external worked example.

1. Independent Replicate Sets

What It Does

By default, the tool treats every well belonging to the same sample/unknown preparation within one assay run as a single pooled set: one corrected mean, one interpolated concentration, one relative potency figure. The optional replicate-set setting instead lets a sample be prepared and plated as several independent physical determinations within the same run — each with its own plates and its own reference-well correction — and has the tool compute a separate corrected mean, interpolated concentration and relative potency for every one of those determinations individually, all against the same shared standard curve fitted once from the standard's own plates.

The independent per-determination results are then combined into a single averaged relative potency and estimated potency for the sample, together with a confidence interval built from the spread between the determinations themselves, and a %RSD (relative standard deviation) suitability check comparing that spread against a limit.

Regulatory and Statistical Background

United States Pharmacopeia, General Chapter <81> — "Antibiotics — Microbial Assays" and general microbial-assay practice call for more than one independent determination of an unknown sample's potency within a single assay run wherever practical, precisely because a single pooled set cannot distinguish ordinary well-to-well noise (already captured by the base method's own confidence interval) from a separate, larger source of variability that only shows up between independently prepared plate sets — for example, small differences introduced during separate dilution preparations, different agar pours, or plates incubated in different positions. A %RSD check across independent replicate determinations is the standard way this second source of variability is monitored and reported in GMP quality-control practice for this assay family.

The Mathematics

Each replicate determination is scored using exactly the same corrected-mean and interpolation formulas as the base (single-set) method — nothing about the per-determination calculation changes:

corrected mean (per replicate set) = test-well mean - (reference-well mean - correction point)

relative potency (per replicate set) = 100 x interpolated concentration / (nominal) reference dose

The n independent relative-potency results are then averaged on the log scale — the same convention used elsewhere in this tool family for combining independent determinations — and a confidence interval is built from their own observed spread, rather than from the well-level pooled variance used for a single set's own interval:

mean(log relative potency) = average of log(relative potency) across the n replicate sets

SD(log relative potency) = sample standard deviation of the same n values (n-1 divisor)

t-critical = Student's t value at (n - 1) degrees of freedom

combined relative potency = exp( mean(log relative potency) )

confidence interval = exp( mean(log relative potency) +/- t-critical x SD(log relative potency) / sqrt(n) )

%RSD = 100 x SD(relative potency) / mean(relative potency), reported directly on the original (not log) scale as the suitability metric

Reasoning: Why This Setting Exists, and When to Use It

Turning this setting on does not change how any individual determination is scored — it only changes how many determinations of the same sample the tool is told to expect, and adds the between-determination summary on top. Where a laboratory's procedure calls for only one physical preparation of the unknown per run, this setting should be left off, and the tool behaves exactly as it always has (a single pooled set, with its confidence interval built from well-level pure-error variance, as before). Where the procedure calls for two or more independent preparations — the more common case for routine antibiotic or microbial potency release testing — this setting surfaces the between-preparation precision check that a paper worksheet would normally compute by hand, without requiring the analyst to run the tool separately for each determination and reconcile the results themselves.

What Changes in the Output, and What Does Not

Affected:

• The potency results table gains a "Replicate Set" column identifying which physical determination each row belongs to
• A new replicate-summary table appears, showing the averaged relative potency, its confidence interval, and %RSD across the replicate sets, with a Pass/Fail suitability result

Not affected:

• The standard curve fit (correction point, intercept, slope, r-squared) — computed once from the standard's own plates, shared unchanged by every replicate set
• Each individual replicate set's own corrected mean, interpolated concentration, and within-set confidence interval — computed identically to the single-set (default) case
• Any assay run where the setting is left unset, which continues to behave exactly as before

Where to Find This Capability

On the Cylinder-Plate 5+1 Assay tool's input form, an optional column selector lets a "Replicate" (or similarly named) column be mapped in, identifying which independent determination each sample well belongs to (e.g. "Set-1", "Set-2", "Set-3"). Standard preparation rows do not need a value in this column. When the column is left unmapped, the tool falls back to its original single-set behavior.

Verification (Internal Consistency Check)

No published worked example using multiple independent replicate sets for this design was located, so this setting was checked for internal consistency instead: the same real, published USP <81> cloxacillin dataset used to verify this tool's core calculation (see the accompanying benchmark note) was re-run with its single test sample's three plates reassigned as three separate one-plate replicate sets rather than one pooled three-plate set.

The tool correctly scored each set independently against the shared standard curve, then combined the three relative-potency figures into a mean of 95.37% with a 95% confidence interval of 84.23%–107.81% and a %RSD of 4.91% (Pass, well under the default 10% limit). As a sanity check, this combined figure is close to the pooled-set result obtained from the same 9 wells treated as a single set in the main benchmark (95.29%), as expected since both are legitimate ways of summarizing the same underlying well readings.

References:

• United States Pharmacopeia, General Chapter <81> — "Antibiotics — Microbial Assays".
• European Pharmacopoeia (Ph. Eur.), Chapter 5.3 — "Statistical Analysis of Results of Biological Assays and Tests".
• Verification dataset: same source as the core Cylinder-Plate 5+1 benchmark below (Stegmann Systems' PLA 3.0 published example report, Document-1994). 

2. Nominal vs. Actual Reference Dose

What It Does

The base method uses a single number — the reference dose — for two different roles at once: it is both the actual concentration physically prepared for the standard's reference (S3) wells, which anchors the standard curve on the log-dose axis, and the value that every sample's interpolated concentration is divided by to produce a relative potency percentage. The optional nominal-reference-dose setting splits these two roles apart, letting the standard curve be anchored on the concentration that was actually prepared, while every reported relative potency and %Assay figure is still expressed as a percentage of the standard's nominal, label-stated target concentration.

Regulatory and Statistical Background

A laboratory's working reference standard is routinely assigned an actual or corrected potency — established against an official or international reference standard and recorded on its certificate of analysis or label — that can differ slightly from its nominal target value (for example, a standard nominally targeted at a round concentration but with a certified actual potency a few percent away from that target). United States Pharmacopeia, General Chapter <111> — "Design and Analysis of Biological Assays" and European Pharmacopoeia (Ph. Eur.), Chapter 5.3 — "Statistical Analysis of Results of Biological Assays and Tests" both require the standard curve itself to be built from the actual prepared concentrations, since that is what determines the true relationship between dose and response; separately, potency results are conventionally still reported relative to the standard's nominal or label value, so that results remain comparable across different lots or certificates of the same reference material issued over time.

The Mathematics

The regression that fits the standard curve is completely unaffected by this setting — it is always anchored using the actual prepared reference dose, exactly as before:

standard curve reference point = ( log(actual reference dose), correction point )  — unchanged

Only the final conversion from an interpolated concentration to a relative-potency percentage changes, switching its divisor from the actual reference dose to the nominal reference dose:

relative potency (%) = 100 x interpolated concentration / nominal reference dose

and the reported confidence interval bounds are shifted by the same ratio, so that the interval remains expressed on the same (nominal) percentage basis as the point estimate:

log-shift = log(actual reference dose / nominal reference dose)

shifted CI bound = original log CI bound + log-shift, then converted back to a percentage

When the nominal reference dose is left unset, it defaults to being equal to the actual reference dose, the log-shift becomes exactly zero, and every output is identical to the tool's original (single-dose) behavior — verified explicitly below.

Reasoning: Why This Setting Exists, and When to Use It

Before this setting existed, a laboratory whose reference standard's actual and nominal potencies genuinely differed had no way to keep both pieces of information distinct inside a single run of the tool — using the actual value everywhere would anchor the curve correctly but report %Assay against the wrong basis, while using the nominal value everywhere would report %Assay on the right basis but subtly miscalibrate the standard curve itself, since the curve must be built from what was actually weighed and diluted on the day of the assay. This setting removes that trade-off entirely.

For the overwhelming majority of runs, where the reference standard's actual prepared concentration and its nominal/label target coincide, this setting should be left unset and has no effect. It should be set explicitly only when the reference standard in active use carries a documented actual potency, distinct from its nominal target, that the laboratory's procedure requires %Assay to be reported against.

What Changes in the Output, and What Does Not

Affected:

• Relative Potency (%) and its confidence interval bounds for every sample — rescaled by the ratio of actual to nominal reference dose

Not affected:

• The standard curve itself: correction point, intercept, slope, r-squared, and every corrected mean
• The Interpolated Concentration figure, which is always reported in the same absolute units as the dose column, independent of this setting
• Estimated Potency, which is derived from Relative Potency and the assumed test potency exactly as before, just using the (now nominal-based) percentage

Where to Find This Capability

On the Cylinder-Plate 5+1 Assay tool's input form, an optional field lets the standard's nominal/label reference concentration be entered separately from the Reference Dose field (the standard's actual prepared concentration). Leaving it blank keeps the tool's original single-value behavior.

Verification (Internal Consistency Check)

This setting was checked for internal consistency against the same real, published USP <81> cloxacillin dataset used to verify this tool's core calculation.

Leaving the setting unset and explicitly setting it equal to the actual reference dose produced identical results in every output, confirming the setting is fully backward compatible. Setting a deliberately different nominal value rescaled the relative potency and its confidence interval by exactly the expected ratio (5.0 / 5.1 = 0.98039; 95.2947% x 0.98039 = 93.4262%, matching the computed 93.4261% to rounding) while leaving the standard curve's own correction point, intercept, and slope completely unchanged, confirming the two roles have been correctly separated.

References:

• United States Pharmacopeia, General Chapter <111> — "Design and Analysis of Biological Assays".
• European Pharmacopoeia (Ph. Eur.), Chapter 5.3 — "Statistical Analysis of Results of Biological Assays and Tests".
• Verification dataset: Stegmann Systems' PLA 3.0 published example report (Document-1994, "PLA_3.0.7_Example_Cylinder-Plate_Assay_USP81_Standard_Report.pdf") — https://www.bioassay.de/fileadmin/user_upload/Webseiten-Landingpages/Bioassay.de/Analytical_methodes/Experiments/Cylinder-plate_assays/PLA_3.0.7_Example_Cylinder-Plate_Assay_USP81_Standard_Report.pdf