📚 Part of the TILA-278 Regulatory Dossier — Reader's Guide. This article shows the live document; edits to the source appear here automatically.

---

Bioanalytical Method Validation (TILA-278)

Document ID: BIOANALYTICAL-001
Version: 1.0
Change History: 1.0 — Initial issue.
Standard(s): ICH M10

Bioanalytical Method Validation — TILA-278

Validation of the bioanalytical assays for TILA-278 in biological matrices (and the anti-drug-antibody assays), covering selectivity, calibration, accuracy and precision, stability, and (for the ligand-binding/immunogenicity assays) the tiered strategy. ICH M10.

1. Scope and Regulatory Framework

This report documents the validation of the bioanalytical methods used to support the clinical pharmacology and immunogenicity characterisation of TILA-278 (Virtual Biopharma Inc.), a humanised IgG1 bispecific monoclonal antibody in which one binding arm antagonises TL1A (TNFSF15) and the second agonises the interleukin-22 receptor (IL-22RA1/IL-10RB) complex. The methods described here generated the concentration and anti-drug-antibody (ADA) data reported in the human pharmacokinetic study report (CLINPHARM-002) and the integrated immunogenicity summary (IMMUNO-001) for the pivotal Phase 2b induction study TILA278-201 in adults with moderate-to-severe ulcerative colitis (UC).

Three method classes are covered:

• a pharmacokinetic (PK) ligand-binding assay for quantitation of TILA-278 in serum;
• a tiered anti-drug-antibody (ADA) assay (screening → confirmation → titre); and
• domain-resolved neutralising-antibody (NAb) assays addressing each functional arm independently.

The PK ligand-binding assay was validated as a full validation in accordance with ICH M10, Bioanalytical Method Validation and Study Sample Analysis. The immunogenicity assays were validated under a risk-based plan aligned with the FDA guidance Immunogenicity Testing of Therapeutic Protein Products — Developing and Validating Assays for Anti-Drug Antibody Detection (January 2019), the EMA immunogenicity guidelines, and the statistical cut-point methodology of the published bioanalytical consensus literature (Shankar/Devanarayan) and USP General Chapters <1106>/<1106.1>; ICH M10 principles for selectivity, matrix interference, stability and study-sample acceptance were applied to these ligand-binding formats where scientifically applicable. Nonclinical exposure and immunogenicity assays supporting the cynomolgus monkey programme — cynomolgus monkey being the sole pharmacologically relevant toxicology species for this molecule under ICH S6(R1) — were validated as species-specific counterparts and are summarised in Section 5.

A defining analytical challenge for this programme is the bispecific architecture: a validated method must confirm that the measured species retains both binding functionalities, must resolve which arm an immune response is directed against, and must remain interpretable in the presence of target-mediated drug disposition (TMDD) driven by both a soluble/membrane TL1A antigen sink and a membrane IL-22R sink. The assay designs below were selected specifically to meet these requirements.

2. Reference Standards and Critical Reagents

The reference standard used for calibration and quality-control (QC) preparation was TILA-278 drug substance from the Chinese hamster ovary (CHO) cell-culture process (Protein A capture followed by polishing chromatography), characterised for content, aggregate level, charge and glycosylation, and traceable by lot to the material used in the clinical and nonclinical studies. Reference-standard and QC lots were bridged whenever the assigned lot changed, with acceptance requiring recovery of the incoming lot within the established accuracy limits against the outgoing lot.

Critical reagents were characterised, lot-controlled, and stability-monitored, with pre-defined lot-to-lot bridging acceptance criteria to guard against reagent drift over the analytical campaign:

• PK assay — a recombinant TL1A-based capture reagent selective for the anti-TL1A paratope and an anti-idiotype detection reagent directed at the IL-22R paratope, together conferring selectivity for the intact bispecific molecule (a molecule lacking either functional arm is not co-captured and detected).
• ADA assay — biotinylated TILA-278 (capture) and SULFO-TAG–labelled TILA-278 (detection) for the bridging format; conjugation ratios were controlled and each new conjugate lot was qualified against the positive control.
• NAb assays — immobilised recombinant human TL1A (competitive ligand-binding format) and an IL-22R–expressing STAT3 reporter cell line (cell-based format), together with recombinant target and single-arm immunodepletion reagents used for domain attribution.
• Positive controls — affinity-purified anti-TILA-278 antibody preparations, including arm-specific preparations used to characterise and qualify the NAb assays; a low positive control at 100 ng/mL anchored sensitivity and drug-tolerance assessments.

3. Pharmacokinetic Assay — Quantitation of Serum TILA-278

3.1 Format and rationale

Total serum TILA-278 was measured by a validated sandwich enzyme-linked immunosorbent assay (ELISA). The recombinant capture reagent engages the anti-TL1A arm and the anti-idiotype detection reagent engages the IL-22R arm, so that only molecules presenting both intact paratopes generate signal. This dual-arm capture/detection design was chosen deliberately over a generic anti-human-Fc format because it reports on structurally intact, dual-binding-competent bispecific and excludes single-arm degradation products, providing a directly interpretable exposure measure for the popPK and exposure-response analyses.

3.2 Calibration and quantitation range

Calibration used the reference standard spiked into pooled blank human serum and diluted to the minimum required dilution (MRD) of 1:10, with a four-parameter logistic (weighted) regression fitted through at least six non-zero calibrator levels plus anchor points and a blank. Validated performance was as follows:

Parameter	Value
Assay format	Sandwich ELISA; recombinant TL1A capture (anti-TL1A arm) + anti-idiotype detection (IL-22R arm)
Analyte measured	Intact bispecific TILA-278 (total)
Minimum required dilution (MRD)	1:10
Calibration model	Four-parameter logistic (weighted); ≥6 non-zero calibrators + anchor points
Lower limit of quantitation (LLOQ)	0.100 µg/mL
Upper limit of quantitation (ULOQ)	30.0 µg/mL (higher concentrations by validated dilution)
Calibration range	0.100–30.0 µg/mL
Inter-assay accuracy (%bias)	−6.8% to +7.4%
Inter-assay precision (%CV)	≤ 11.2%
Selectivity / matrix effect	No significant interference from normal or UC serum

Per-run calibration acceptance followed ICH M10: at least 75% of calibrators (and a minimum of six) within ±20% of nominal (±25% at the LLOQ and ULOQ).

3.3 Accuracy and precision

Accuracy and precision were established across a minimum of six independent validation runs by more than one analyst, using QC samples at the LLOQ (0.100 µg/mL) and at low, mid and high concentrations spanning the calibration range, plus an above-ULOQ dilution QC. Inter-assay accuracy ranged from −6.8% to +7.4% bias and inter-assay precision was ≤11.2% CV, within the ICH M10 acceptance limits for a ligand-binding assay of ±20% accuracy and ≤20% precision (±25% and ≤25% at the LLOQ), with total error within the 30% limit (40% at the LLOQ). In-study run acceptance applied the ligand-binding QC rule: at least 67% of all QC results, and at least 50% at each concentration level, within ±20% of nominal (±25% at the LLOQ).

3.4 Selectivity, specificity and matrix effects

Selectivity was demonstrated in at least ten individual lots of normal human serum and in individual UC (disease-state) serum, each evaluated at the LLOQ and at a high QC level, with recovery within ±20% (±25% at the LLOQ) in at least 80% of lots. Additional matrix conditions relevant to the study population were assessed, including haemolysed and lipaemic serum and sera containing concomitant medications common to the UC population (corticosteroids, immunomodulators, aminosalicylates), together with a rheumatoid-factor challenge. Specificity was supported by the anti-idiotype detection reagent and confirmed by the absence of measurable cross-reactivity with structurally or mechanistically related molecules (endogenous TL1A, IL-22 and the IL-22R/DR3 axis), so that neither endogenous ligand generated false signal in the sandwich format.

3.5 Dilutional linearity, parallelism and high-dose hook

Dilutional linearity was demonstrated for samples above the ULOQ by spiking above the calibration range and diluting into range, with back-calculated recovery within ±20% of nominal; this supports quantitation of the peak concentrations observed on the High regimen after dilution into the calibrated range. The absence of a high-dose (prozone) hook effect was verified up to 1000 µg/mL, well above observed peak concentrations. Parallelism was assessed by serial dilution of high-concentration incurred (study) samples; back-calculated concentrations across dilutions agreed within the pre-specified limit (≤30% CV), confirming that the calibrator/matrix surrogate behaves like authentic circulating analyte.

3.6 Stability

Analyte stability in serum was established under the conditions encountered from collection through analysis: bench-top stability at room temperature and at 2–8°C, at least three freeze–thaw cycles between the frozen storage condition and room temperature, and long-term frozen storage of at least 12 months at ≤−70°C (with an interim −20°C condition). Stock and working reference-standard and critical-reagent solution stability were likewise established, and all study samples were analysed within the demonstrated stability windows.

3.7 Target interference under target-mediated disposition

Because TILA-278 exhibits TMDD from both arms, the potential for circulating target to interfere with quantitation was evaluated explicitly. Recombinant soluble TL1A was spiked across the physiological-to-elevated range against fixed analyte concentrations; measured TILA-278 remained within accuracy limits, confirming that the sandwich format is not materially biased by the soluble TL1A antigen sink and reports total intact bispecific rather than a sink-depleted fraction. The IL-22R arm engages a membrane-anchored receptor that does not circulate at interfering soluble concentrations, so target interference on the detection side is not expected and was confirmed to be absent. These results underpin the interpretation of the trough and intensive PK profiles and of the saturable (Michaelis-Menten) clearance component described in CLINPHARM-002.

4. Immunogenicity Assays — Tiered ADA Strategy

4.1 Overview and platform

Binding ADA were measured by a validated three-tier scheme (screening → confirmation → titre) on an electrochemiluminescence (ECL) bridging platform (Meso Scale Discovery, MSD), using biotinylated TILA-278 as capture and SULFO-TAG–labelled TILA-278 as detection. The homogeneous bridging format detects IgG and IgM isotypes and is agnostic to which arm the antibody engages. Acid dissociation was incorporated in sample pre-treatment to release drug-bound ADA and thereby improve drug tolerance in the presence of circulating TILA-278. Confirmed-positive samples were then characterised for neutralising activity by the domain-resolved assays in Section 4.5.

4.2 Tier 1 — screening

The screening cut point was established statistically from drug-naïve UC and healthy-donor serum to yield a nominal 5.0% false-positive rate (95th percentile of the normalised signal distribution), applied as a floating (per-plate) cut point. Assay sensitivity, determined with the affinity-purified positive control, was 22.8 ng/mL, comfortably exceeding the ≤100 ng/mL expectation for a therapeutic-protein ADA assay. Drug tolerance at the 100 ng/mL low positive control was 25 µg/mL, and no signal-suppressing hook was observed to 500 µg/mL, so that screening reliably interrogates samples at the pre-dose (trough) concentrations encountered across the sampling schedule.

4.3 Tier 2 — confirmation

Screening-positive samples were confirmed by competitive inhibition (drug competition/immunodepletion): each sample was re-assayed after pre-incubation with excess unlabelled TILA-278, and drug-specific signal suppression above the confirmatory cut point — set to a 1.0% false-positive rate (99th percentile of inhibition in drug-naïve samples) — designated the sample confirmed ADA-positive. Drug tolerance in the confirmatory tier was likewise 25 µg/mL at the 100 ng/mL positive control.

4.4 Tier 3 — titre

Confirmed-positive samples were serially diluted and re-assayed against a titre cut point corresponding to a 0.1% false-positive rate; titre was reported as the reciprocal of the highest dilution yielding a response at or above the titre cut point and summarised in ordinal low/moderate/high categories, with titre-assay reproducibility within ≤1 titre step.

4.5 Neutralising-antibody assays — domain-resolved

Reflecting the dual mechanism (TL1A antagonism plus IL-22R agonism), all confirmed ADA-positive samples were characterised for neutralising activity using two orthogonal, arm-specific assays:

• Anti-TL1A arm (antagonist) — a competitive ligand-binding (CLB) assay measuring inhibition of TILA-278 binding to immobilised recombinant TL1A; neutralising ADA reduce the drug–TL1A signal below the NAb cut point. Validated sensitivity was 478 ng/mL and drug tolerance 20 µg/mL, with no hook to 250 µg/mL and a 1.0% cut-point false-positive target.
• IL-22R arm (agonist) — a cell-based bioassay in an IL-22R–expressing reporter line measuring TILA-278–driven STAT3-pathway activation; neutralising ADA suppress the agonist-induced reporter signal. Validated sensitivity was 642 ng/mL and drug tolerance 15 µg/mL, with no hook to 250 µg/mL and a 1.0% cut-point false-positive target. A CLB confirmatory format (inhibition of TILA-278 binding to recombinant IL-22R) was available for samples with matrix interference in the cell-based readout.

To preserve neutralising-antibody detection in the presence of circulating drug, NAb samples underwent acid dissociation together with a target-capture bead-extraction step, achieving drug tolerance in excess of the observed pre-dose trough concentrations. Where sample volume permitted, immunodepletion with single-arm reagents attributed neutralising activity to the anti-TL1A paratope, the IL-22R paratope, or both.

4.6 Domain specificity of binding ADA

Because the bispecific presents a native humanised IgG1 framework alongside a non-native geometry and a novel IL-22R agonist epitope, confirmed binding ADA were further resolved for domain specificity — directed at the anti-TL1A paratope, the IL-22R paratope, both, or framework/linker regions — using single-arm competition/immunodepletion reagents. This resolution supports the interpretation of neutralising results and of any potential interaction with endogenous IL-22 biology.

4.7 Validation summary

All parameters met the pre-defined acceptance criteria of the risk-based validation plan.

Table 4-1. ADA and NAb assay validation summary.

Parameter	Screening (ECL bridging)	Confirmatory	Titre	NAb anti-TL1A (CLB)	NAb IL-22R (cell-based)
Cut point (false-positive target)	5.0% (95th pct)	1.0% (99th pct)	Titre cut point (0.1%)	1.0%	1.0%
Sensitivity (affinity-purified positive control)	22.8 ng/mL	—	—	478 ng/mL	642 ng/mL
Drug tolerance (at 100 ng/mL PC)	25 µg/mL	25 µg/mL	—	20 µg/mL	15 µg/mL
Intra-/inter-assay precision (CV)	<12% / <18%	<12% / <18%	≤1 titre step	<15% / <20%	<15% / <20%
Hook effect	None to 500 µg/mL	—	—	None to 250 µg/mL	None to 250 µg/mL
Selectivity/matrix interference	Acceptable (individual and disease-state sera)	Acceptable	Acceptable	Acceptable	Acceptable

Drug tolerance = highest TILA-278 concentration at which the 100 ng/mL positive control remains detectable. Selectivity was demonstrated in individual and disease-state (UC) sera, including evaluation of rheumatoid factor and disease-associated autoantibody interference. NAb sample pre-treatment (acid dissociation plus target-capture bead extraction) provided drug tolerance exceeding the highest observed pre-dose trough concentration.

The screening sensitivity of 22.8 ng/mL and drug tolerance to 25 µg/mL exceed the ≤100 ng/mL sensitivity expectation and cover the pre-dose (trough) drug concentrations observed across the sampling schedule, supporting reliable ADA detection; the arm-resolved NAb assays likewise retained detection above the observed troughs.

5. Cross-Validation, Cross-Species and Study-Sample Analysis

Nonclinical (cynomolgus) counterparts. Species-specific PK and immunogenicity methods were validated in cynomolgus monkey serum to support the toxicology and toxicokinetic programme, cynomolgus monkey being the sole pharmacologically relevant species for both binding arms under ICH S6(R1). Method equivalence between the human and cynomolgus formats was established by cross-validation of shared critical reagents and calibration behaviour. Consistent with class understanding and ICH S6(R1), rodent immunogenicity data are not predictive of human ADA responses and were not used to characterise clinical immunogenicity; the clinical assays above provide the definitive human characterisation.

Study-sample acceptance and reanalysis. Analytical runs were accepted per the ICH M10 ligand-binding criteria (Section 3.2–3.3), and study samples with results above the ULOQ were re-assayed after validated dilution into the calibration range. Reasons for any reanalysis were documented and controlled.

Incurred sample reanalysis (ISR). ISR was performed on a pre-defined subset of study samples for the PK assay; agreement met the ligand-binding acceptance criterion of at least 67% of repeats within ±30% of the original result, confirming reproducibility of TILA-278 quantitation in authentic incurred serum. Reproducibility of the ADA/NAb determinations was supported by the per-tier precision in Table 4-1 and by consistent cut-point performance across the analytical campaign.

6. Conclusion

The bioanalytical methods supporting the TILA-278 clinical pharmacology and immunogenicity programme were validated to their intended purpose. The PK sandwich ELISA specifically quantifies the intact bispecific over a 0.100–30.0 µg/mL range with inter-assay accuracy of −6.8% to +7.4% and precision ≤11.2%, is free of soluble-TL1A target interference, and met ICH M10 selectivity, dilutional-linearity, parallelism, stability and ISR expectations. The tiered ADA scheme (screening 5.0%, confirmatory 1.0%, titre 0.1% cut points) with 22.8 ng/mL sensitivity and 25 µg/mL drug tolerance, complemented by domain-resolved NAb assays against the anti-TL1A (478 ng/mL) and IL-22R (642 ng/mL) arms, meets current FDA and EMA immunogenicity expectations and reliably supports the exposure, exposure-response and immunogenicity conclusions reported in CLINPHARM-002 and IMMUNO-001. These validated methods are adequate to support the clinical pharmacology data package for the biologics licence application filed under 21 CFR Part 601. ICH M10.

---

Validation parameters are reported to the conventions of ICH M10 and the cited FDA/EMA immunogenicity guidances for a humanised IgG1 bispecific antibody exhibiting target-mediated disposition; assay performance values are consistent with, and were the source of, the bioanalytical parameters referenced in the associated clinical pharmacology (CLINPHARM-002) and immunogenicity (IMMUNO-001) reports.