chapter 02 · the operational layer

BE substrate: CROs, volunteers, units, contracts, conduct.

The operational architecture beneath every BE study. CRO oversight model · healthy-volunteer regimes (registries, screening, washouts) · clinical-pharmacology unit infrastructure · sponsor-CRO contracting · the five-phase study conduct sequence · the bioanalytical bridge. The substrate iFeed runs on.

/ 01

CRO oversight model.

5 contracting patterns

The BE CRO is not a single homogeneous category. The contracting pattern determines who owns the protocol, who owns the audit trail, who is liable when ANVISA inspects, who answers a 483 observation. Five patterns dominate the 2026 market — each with distinct sponsor exposure, distinct cost, distinct time-to-readiness.

/ Pattern 01

Full-service CRO.

End-to-end: protocol, ethics filing, recruitment, dosing, sampling, bioanalysis, statistics, CSR. Sponsor signs off; CRO owns conduct. Default for first-time generic sponsors. Higher cost; lower sponsor operational load.

/ Pattern 02

Functional service · FSP.

Sponsor retains protocol authorship and CSR; CRO provides conduct phases (clinical conduct, bioanalysis). Common for repeat-generic sponsors with internal regulatory teams.

/ Pattern 03

Split-CRO · clinical / bioanalytical.

Clinical conduct (CPU dosing) at one CRO; bioanalysis at another. Common for sponsors leveraging specialised LC-MS/MS labs. Cross-CRO data integration risk.

/ Pattern 04

Captive CPU.

Sponsor-owned clinical-pharmacology unit. Increasingly rare; used by largest generics (Sun Pharma, Cipla) for India-domestic studies. Highest control, highest fixed-cost overhead.

/ Pattern 05

Hybrid · academic-CPU.

University clinical pharmacology unit conducts the dosing and sampling; commercial CRO handles bioanalysis. Common for biosimilar bridging studies and rare-disease BE.

/ 02

Healthy-volunteer regimes.

Registries, screening, washouts

The healthy volunteer is the unit of work for a BE study. The 2024 ANVISA Rules made this layer regulated for the first time at the volunteer-registry level — meaning a single volunteer cannot enrol in overlapping BE studies within the washout window. The complete landscape:

ANVISA 2024
Mandatory healthy-volunteer registry. National-level registry. No overlapping enrolment within washout. Brazilian-distinctive operational rule following 2022 RDC 742.
2024
FDA 21 CFR 320
Healthy adult volunteer default. Sponsor responsibility to confirm no concurrent participation. No central registry; relies on sponsor screening.
1977
EMA 2010
EMA Guideline §3.2. Healthy volunteer screening, exclusion criteria, washout calculation. National-level guidance; some Member States operate registries (Belgium, France).
2010
CDSCO India
Schedule Y / 2018 BE GL. Healthy volunteer guidance under Drugs & Cosmetics Act. CRO-level recordkeeping; inter-CRO sharing under sponsor responsibility.
2018
WHO TRS 992
Annex 7 §11. Healthy volunteer protection in PQ-bound BE studies. Aligns with ICH E6 GCP principles.
aligned
PMDA Japan
PMDA BE Guideline. Healthy Japanese volunteers; ethnicity considerations explicit for cross-region bridging studies.
aligned
/ 03

Clinical-pharmacology unit infrastructure.

5 components

The CPU is where the BE study actually happens — the place a healthy volunteer arrives at 06:00, gets dosed at 08:00, and stays through 24/48/72 hours of sampling. Five infrastructure components. Each one a regulator inspection target. Each one a cause of failure if absent.

/ 01
Screening rooms.

Vital signs, ECG, blood draw for safety labs. 7-day screening window typical; pregnancy testing day-of for women of child-bearing potential.

/ 02
Dosing bay.

Witnessed administration. IMP storage at room temp / 2-8 °C. Mouth check post-dosing. Drinking water restrictions for ±1h around dose.

/ 03
Sampling bay.

Indwelling cannula. Pre-dose + 18-22 timepoint sampling. Plasma separation within 30 min of collection. Storage at -70 to -80 °C.

/ 04
Crossover housing.

Period 1 confinement 24-48h. Washout-window release. Period 2 readmission. Same volunteer, same beds, same meals.

/ 05
Pharmacy & archive.

IMP receipt, accountability ledger, randomisation envelopes. Source data archive (paper + EDC). 15-year retention typical post-marketing authorisation.

/ 04

Sponsor-CRO contracting patterns.

8 frame contracts

The contract is the legal substrate. Once a 483 observation lands or a deficiency letter arrives, the contract determines who pays for the rerun. Eight contract types cover almost all BE work in 2026.

/ Contract 01
Fixed-price study contract.

Single deliverable. CSR + audit-ready dossier. Most common for first-time generics. Risk on CRO for cost overruns.

/ Contract 02
Time-and-materials.

Hourly + reagent costs. Used for novel formulations or unknown screening yield. Risk on sponsor.

/ Contract 03
Per-volunteer.

Common in India CROs. Per-volunteer fee covers screening + dosing + sampling. Bioanalysis costed separately.

/ Contract 04
Master service agreement.

MSA + Work Orders. Used by repeat-generic sponsors filing many ANDAs / generic registrations annually.

/ Contract 05
Risk-share rerun.

If first-pass BE fails, CRO covers rerun cost. Higher initial fee. Used by sponsors with tight ANDA timelines.

/ Contract 06
Pass-through bioanalysis.

CRO subcontracts to specialist BMV lab. Sponsor signs three-way contract; data flows back to CRO for CSR integration.

/ Contract 07
Inspection indemnity.

Specific clause for FDA / EMA / ANVISA inspection findings. Standard since the 2010s; tightened post-Cetero (2011) and Semler (2016).

/ Contract 08
Data ownership clause.

Raw data, audit trail, source-data ownership. Sponsor typically retains; CRO keeps copy under regulator-defined retention.

/ 05

Study conduct phases.

5-phase sequence

The five phases are sequential, with regulator-specified handoffs. Skipping a phase, or running them out of order, is the most common cause of inspection observations during BE conduct — particularly around washout and crossover transitions.

/ Phase 01
Run-in.

7-day screening window. Vital signs, ECG, lab safety screen. Pregnancy test day-of-dose. Confirm no concurrent meds, no recent BE participation. Eligibility lock at -1 day.

/ Phase 02
Dosing.

Period 1. Randomisation envelope opened. Witnessed dose at T=0. Pre-dose blood sample. Post-dose timepoints per protocol (typical 18-22 timepoints over 24-72h).

/ Phase 03
Sampling.

Plasma collection per timepoint. K2-EDTA tubes typical. Centrifuge within 30 min. Aliquot to 2 cryotubes (primary + backup). Storage -70 to -80 °C.

/ Phase 04
Washout.

Inter-period gap. ≥7 half-lives recommended (covers >99% drug clearance). Volunteer released; readmits at end of washout. Concurrent BE participation prohibited.

/ Phase 05
Crossover.

Period 2. Same protocol as Period 1 with crossover treatment per randomisation. Same sampling timepoints. Same safety monitoring. Subject completion at last sample + safety follow-up.

/ 06

The bioanalytical bridge.

Where BE meets BMV

The BE study and the bioanalytical method are joined at the hip. The BE plasma samples generated during conduct must be analysed with a method validated to ICH M10 / FDA BMV / EMA standards before the GMR can be calculated. The handoff between clinical conduct and bioanalysis is where most data-integrity findings emerge in inspection.

/ Cross-domain bridge

BE conduct → BMV analysis → statistics.

Validated method (ICH M10) consumes BE plasma samples. Calibration curve and QCs run with each batch. ISR mandatory at 10% sample reanalysis (or 20% for selected drugs). Failed-batch acceptance criteria pre-defined. The clinical-bioanalytical handoff is regulator-inspected as one workflow.

M10
Sample receipt
Chain-of-custody at receipt. CRO → BMV lab transfer log. Temperature monitoring during transit. Receipt log signed within 24h.
M10 §6.3
Calibration curve
Per-batch calibration. 6 non-zero calibrators, blank, zero. ±15% nominal except LLOQ at ±20%. Same calibrators valid for one batch only.
M10 §3.3
QC samples
4 QC concentration levels. Low (3× LLOQ), Mid, High (~75% ULOQ), and one near-LLOQ. ±15% accuracy / 15% CV precision.
M10 §3.4
ISR
Incurred sample reanalysis. 10% of clinical samples reanalysed. ≥67% within ±20% of original. Mandatory for pivotal BE studies.
M10 §3.5.4
Failed-batch
Acceptance criteria. 75% of QCs and ≥67% per concentration. Reanalysis authority pre-defined per SOP. Documented per study.
M10 §3.5.3
Statistical handoff
Concentration-time data → statistician. Cmax, AUC0-t, AUC0-inf calculation. Log-transform. TOST. 90% CI. Per-protocol analysis set definition pre-locked at SAP.
M13A §6
/ 07

Data integration: clinical ↔ bioanalytical.

EDC, LIMS, statistics

The BE substrate ends in three parallel data streams that must reconcile before submission: clinical EDC (concomitant meds, AE, vitals), bioanalytical LIMS (concentrations per sample), and statistical SAP output (PK parameters + TOST). Reconciliation is where the SDTM / ADaM model becomes load-bearing for ANDA / EU MAA filings.

/ Layer 01

Clinical EDC.

Medidata Rave · Veeva Vault EDC · Oracle Inform. Concomitant meds, AE, vitals, dosing log, randomisation. Source-data verification per CRA visit. Audit-trail per 21 CFR 11.

/ Layer 02

Bioanalytical LIMS.

Watson LIMS · SCIEX OS · Waters MassLynx + iSubmit. Sample inventory, calibration curves, QC results, raw data archive. Part 11-compliant audit trail.

/ Layer 03

Statistical analysis.

SAS 9.4 · Phoenix WinNonlin · R + ncar / nlmixr. PK parameter calculation (Cmax, AUC0-t, AUC0-inf). Log-transform + TOST. SAP-locked outputs.

Reconciliation point 01
Sample list. EDC randomisation table reconciled to BMV LIMS sample receipt. Single mismatched volunteer ID is a Phase-3-stop event.
pre-stats
Reconciliation point 02
Concomitant meds. EDC con-med log reconciled to potential bioanalytical interference. Excluded subjects per per-protocol analysis set rules.
pre-stats
Reconciliation point 03
Concentration-time data lock. BMV LIMS → statistical input dataset. Database lock signed by sponsor + CRO + statistician before TOST.
stats input
Reconciliation point 04
SAP output. Cmax, AUC, GMR, 90% CI. Per-protocol + intent-to-treat sets. Submitted to ANDA / MAA / WHO PQ.
submission