0. Executive Framework

The objective is to establish a science- and risk-based process understanding system that connects:

QTPP → CQAs → CMAs/CPPs → Risk Assessment → DoE → Design Space → Control Strategy → PAT → Process Qualification → CPV → Lifecycle Management

The fundamental principle is:

Quality should be designed into the product and process rather than tested into the finished product.

ICH Q8 establishes the pharmaceutical-development/QbD foundation, Q9 provides the quality-risk-management framework, Q10 establishes the pharmaceutical quality system, Q12 addresses lifecycle management and post-approval changes, Q13 addresses continuous manufacturing, and Q14 addresses analytical procedure development and lifecycle management.

FDA’s PAT framework similarly emphasizes scientific understanding of the process, risk-based control, and innovative process monitoring throughout the product lifecycle.

1. QbD Foundation — QTPP, CQAs and Risk Assessment

1.1 Define the Quality Target Product Profile

The QTPP is the prospective description of the product characteristics required to achieve the intended clinical performance.

QTPP Template

Deliverable

QTPP Document + Product Knowledge Summary

1.2 Identify Critical Quality Attributes

A CQA is a physical, chemical, biological or microbiological property that must remain within an appropriate limit, range or distribution to assure product quality.

Example — Solid Oral Dosage Form

Potential CQAs include:

• Assay
• Content uniformity
• Dissolution
• Related substances
• Degradation products
• Tablet weight
• Hardness
• Friability
• Disintegration
• Moisture
• Appearance
• Microbial quality

Example — Monoclonal Antibody

Potential CQAs include:

• Identity
• Purity
• Aggregates
• Fragments
• Charge variants
• Glycosylation
• Potency
• Protein concentration
• Host-cell proteins
• Residual DNA
• Endotoxin
• Bioburden
• Viral clearance
• Particulates

1.3 CQA Criticality Assessment

Use a structured scoring approach.

Suggested Criticality Score

Severity × Probability × Detectability = Risk Priority Number

Example:

The scoring system should be defined in the organization’s QRM procedure rather than treated as a universal regulatory scoring scale.

1.4 Process Mapping

Create a complete process flow:

Raw Materials → Dispensing → Processing Step 1 → Processing Step 2 → Intermediate → Final Processing → Packaging → Finished Product

For each operation identify:

1. Process input
2. CMA
3. Process parameter
4. CPP
5. Intermediate CQA
6. Final CQA
7. Existing control
8. Proposed PAT
9. Sampling location
10. Sampling frequency
11. Failure mode

Example — OSD

Dispensing → Granulation → Drying → Milling → Lubrication → Compression → Coating → Packing

Example — mAb Downstream

Harvest → Clarification → Protein A → Viral Inactivation → Polishing → Virus Filtration → UF/DF → Final Bulk

1.5 Initial Risk Assessment

Use a combination of:

• Ishikawa/Fishbone
• FMEA
• HACCP
• Fault Tree Analysis
• Cause-and-effect matrix
• Historical manufacturing data
• Prior knowledge
• Scientific literature
• Development studies

FMEA Template

1.6 Build the CQA–CPP–CMA Linkage

A key QbD deliverable is a CQA-CMA-CPP matrix.

The goal is to transform a qualitative risk assessment into experimentally demonstrated process understanding.

2. DoE Strategy and Design Space

2.1 DoE Philosophy

DoE should answer three progressively deeper questions:

Stage 1 — Screening

Which factors matter?

Stage 2 — Characterization

How do important factors affect CQAs?

Stage 3 — Optimization

Where should the process operate to consistently achieve the desired quality?

Avoid changing one factor at a time as the primary development methodology when interactions are scientifically important.

2.2 Screening DoE

Potential approaches:

• Fractional factorial
• Full factorial
• Plackett-Burman
• Definitive screening design
• Custom optimal design

Example

Suppose eight potential CPPs are identified:

• Temperature
• pH
• Mixing speed
• Mixing time
• Feed rate
• Pressure
• Concentration
• Hold time

A screening DoE can determine which factors have meaningful effects on:

• Assay
• Purity
• Dissolution
• Particle size
• Yield
• Aggregation
• Impurity clearance

Screening Deliverables

• Factor selection
• Experimental matrix
• Randomization strategy
• Blocking strategy
• Response definitions
• Statistical model
• ANOVA
• Main-effect plots
• Interaction assessment
• Pareto chart
• Significant-factor list

2.3 Optimization DoE

After screening, select significant factors.

Typical approaches:

Response Surface Methodology

Use when:

• curvature is expected;
• interactions are important;
• optimization is required.

Central Composite Design

Useful for:

• quadratic response models;
• center points;
• estimating curvature;
• broad factor ranges.

Box-Behnken Design

Useful when:

• three or more factors are being optimized;
• extreme combinations are undesirable;
• quadratic modeling is required.

2.4 Example Optimization Model

For a CQA such as dissolution:

Y = β₀ + β₁X₁ + β₂X₂ + β₃X₃ + β₁₂X₁X₂ + β₁₃X₁X₃ + β₂₃X₂X₃ + β₁₁X₁² + β₂₂X₂² + β₃₃X₃²

Where:

• Y = response
• X₁ = compression force
• X₂ = lubricant concentration
• X₃ = granule moisture

Evaluate:

• statistical significance;
• model adequacy;
• residuals;
• lack of fit;
• prediction capability;
• interaction effects;
• practical significance.

2.5 Design Space

The Design Space should represent the multidimensional combination of material attributes and process parameters demonstrated to provide assurance of meeting predefined quality requirements.

Conceptually:

Design Space = Combination of CMAs + CPPs that delivers acceptable CQAs

Example:

Important distinction

PAR — Proven Acceptable Range

Range experimentally demonstrated to provide acceptable quality.

NOR — Normal Operating Range

The routine operating range selected inside the proven acceptable region, generally providing additional operational margin.

Design Space

Can be multidimensional and does not have to be represented as independent limits for every parameter.

2.6 Design Space Verification

Before implementation:

• verify scale dependence;
• evaluate equipment differences;
• assess raw-material variability;
• challenge edge-of-range conditions;
• evaluate worst-case combinations;
• confirm model predictions;
• perform engineering/registration batches where appropriate.

Deliverables

• Design Space Report
• Statistical Model Report
• Edge-of-Space Verification Protocol
• Design Space Verification Report
• Updated Control Strategy
• Regulatory filing content, where applicable

3. PAT Strategy

FDA’s PAT framework promotes process understanding and risk-based implementation of analytical technologies for process monitoring and control.

PAT should not be selected merely because an instrument is technologically advanced.

The fundamental question is:

What process or quality decision will this measurement enable?

3.1 PAT Technology Selection Matrix

3.2 PAT Classification

Inline

Measurement occurs directly within the process stream without removing the sample.

Examples:

• Inline Raman
• Inline NIR
• Inline UV
• Inline conductivity
• Inline pressure
• Inline temperature

Online

Sample is automatically transported to an analyzer.

At-line

Sample is removed and analyzed close to the process.

3.3 PAT Control Hierarchy

A mature PAT program should evolve through:

Monitor → Detect → Predict → Control → Optimize

Example:

Level 1 — Monitoring

NIR measures moisture.

Level 2 — Detection

System identifies abnormal drying trajectory.

Level 3 — Prediction

Chemometric model predicts final moisture.

Level 4 — Control

Drying endpoint automatically terminates.

Level 5 — Optimization

System dynamically adjusts process conditions based on model predictions.

3.4 Chemometric Model Lifecycle

For NIR/Raman or other multivariate PAT:

Step 1 — Define Intended Use

Specify:

• attribute being predicted;
• operating range;
• measurement location;
• sample matrix;
• accuracy requirement;
• decision criteria.

Step 2 — Data Generation

Collect representative samples covering:

• normal variation;
• raw-material variability;
• process extremes;
• scale;
• equipment;
• manufacturing sites where applicable.

Step 3 — Spectral Preprocessing

Potential techniques:

• SNV
• MSC
• derivatives
• smoothing
• baseline correction

Step 4 — Model Development

Possible approaches:

• PLS
• PCR
• PCA
• SIMCA
• multivariate classification
• machine-learning models where scientifically justified

Step 5 — Model Validation

Evaluate:

• RMSEC
• RMSEP
• bias
• precision
• robustness
• specificity
• leverage
• outliers
• prediction uncertainty

FDA has specific guidance addressing development and submission of NIR analytical procedures using chemometric models.

Step 6 — Model Deployment

Control:

• software;
• instrument configuration;
• model version;
• cybersecurity;
• data integrity;
• audit trail;
• access control.

Step 7 — Model Monitoring

Monitor:

• spectral drift;
• instrument drift;
• prediction residuals;
• applicability domain;
• reference-method correlation;
• abnormal spectra.

Step 8 — Model Maintenance

Establish:

Model Change Control → Impact Assessment → Recalibration/Update → Verification → Approval → Deployment

3.5 PAT and Real-Time Release Testing

Where scientifically justified, PAT can support Real-Time Release Testing (RTRT).

The fundamental chain is:

PAT Measurement → Validated Model → CQA Prediction → Acceptance Criterion → Batch Disposition

This requires strong scientific evidence demonstrating that the PAT measurement/model provides reliable information about the relevant quality attribute.

ICH Q14 specifically includes considerations for multivariate analytical procedures and real-time release testing.

4. Control Strategy Development

The final QbD output should be a scientifically justified Control Strategy.

4.1 Control Strategy Components

Include:

Material Controls

• API attributes
• Excipient properties
• Raw-material variability
• Supplier controls

Process Controls

• CPP limits
• IPCs
• PAT
• alarms
• interlocks
• automated controls

Facility/Equipment Controls

• equipment capability;
• calibration;
• environmental controls;
• utilities;
• cleaning;
• maintenance.

Analytical Controls

• release testing;
• stability testing;
• PAT;
• reference methods;
• sampling strategy.

Procedural Controls

• SOPs;
• batch records;
• training;
• deviation management;
• change control.

5. Process Validation and Technology Transfer

FDA’s process-validation framework treats validation as a lifecycle activity encompassing process design, process qualification and continued process verification.

Stage 1 — Process Design

Demonstrate:

• scientific understanding;
• QTPP;
• CQAs;
• risk assessment;
• DoE;
• Design Space;
• control strategy.

Stage 2 — Process Qualification

Demonstrate:

• facility readiness;
• equipment qualification;
• utilities;
• manufacturing reproducibility;
• process performance;
• operator readiness;
• analytical readiness.

Stage 3 — Continued Process Verification

Demonstrate:

• ongoing state of control;
• process capability;
• trend stability;
• CQA consistency;
• variability reduction;
• timely detection of emerging risks.

6. Continued Process Verification — CPV

FDA describes CPV as ongoing collection and evaluation of process and product data to ensure the process remains in a state of control.

CPV should not be treated as simply generating annual trend charts.

It should function as a process-health surveillance system.

6.1 CPV Data Architecture

Collect data from:

Material

• supplier;
• lot;
• particle size;
• moisture;
• potency;
• density;
• critical excipient attributes.

Process

• CPPs;
• equipment;
• operator/shift;
• environmental conditions;
• batch duration;
• process interruptions.

Product

• CQAs;
• IPCs;
• release results;
• stability;
• deviations;
• complaints.

PAT

• spectra;
• predicted CQAs;
• endpoint data;
• sensor trends;
• model diagnostics.

6.2 SPC Strategy

Use different statistical tools for different signals.

Shewhart Charts

Best for detecting large, sudden shifts.

I-MR Charts

Useful for individual batch measurements.

X-bar/R or X-bar/S Charts

Useful where rational subgroups exist.

EWMA

Useful for detecting small persistent shifts.

CUSUM

Useful for detecting gradual process drift.

6.3 Process Capability

Use:

Cp / Cpk

when the process is statistically stable and the specification limits are meaningful.

Use:

Pp / Ppk

to describe longer-term observed performance.

Example:

Cpk = minimum[(USL − μ)/(3σ), (μ − LSL)/(3σ)]

Do not interpret a high Cpk as proof of process control if the process is unstable.

6.4 Suggested CPV Classification

Thresholds should be scientifically justified and defined in the approved CPV procedure.

6.5 Multivariate CPV

For highly interconnected processes, univariate charts may miss emerging problems.

Consider:

• PCA
• PLS
• Mahalanobis distance
• Hotelling’s T²
• SPE/Q residuals
• multivariate control charts
• multivariate batch fingerprinting

Example:

20 CPPs + 10 CQAs → Multivariate Process Health Index

This can provide earlier detection of complex process drift than monitoring each parameter independently.

7. Feedback and Feedforward Control

A mature QbD system should create a closed-loop knowledge architecture.

Feedforward

Material Attribute → Predictive Model → Process Adjustment

Example:

Higher API moisture detected → adjust granulation/drying parameters.

Feedback

Process Measurement → CQA Prediction → Process Correction

Example:

PAT predicts moisture above target → drying conditions automatically adjusted.

CPV Feedback

Commercial Data → Trend Detection → Investigation → CAPA/Change → Updated Control Strategy

This is the foundation of continual improvement.

8. ICH Q12 Lifecycle Management

ICH Q12 provides a framework intended to make post-approval CMC change management more predictable and efficient while supporting continual improvement.

The lifecycle architecture should therefore distinguish:

Established/Registered Elements

Parameters or attributes requiring regulatory consideration according to the applicable filing and jurisdiction.

Operational Controls

Routine manufacturing controls managed within the pharmaceutical quality system.

Knowledge Management

Accumulated scientific and manufacturing knowledge supporting future decisions.

Change Management

Every proposed change should undergo:

Change → QRM → Impact on CQA/CPP/Design Space → Regulatory Assessment → Implementation → Effectiveness Verification

9. Knowledge Management System

Create a living Process Knowledge Repository.

Maintain:

• QTPP;
• CQA list;
• CMA list;
• CPP list;
• risk assessments;
• DoE datasets;
• statistical models;
• Design Space;
• PAT models;
• CPV data;
• deviations;
• CAPA;
• complaints;
• stability;
• process changes;
• validation reports;
• technology-transfer knowledge.

This converts isolated development studies into a lifecycle process knowledge base.

10. Recommended Lifecycle Implementation Roadmap

Phase 1 — Product Understanding

Activities

• QTPP definition
• CQA identification
• prior knowledge review
• product characterization

Deliverables

• QTPP
• CQA register
• Product Knowledge Report

Phase 2 — Process Understanding

Activities

• process mapping
• CMA identification
• CPP identification
• initial FMEA
• Ishikawa analysis

Deliverables

• Process Flow Diagram
• CQA-CMA-CPP Matrix
• Initial Risk Assessment

Phase 3 — Experimental Development

Activities

• screening DoE
• optimization DoE
• interaction studies
• scale-up studies

Deliverables

• DoE Protocol
• DoE Report
• Statistical Model
• Significant-Factor Matrix

Phase 4 — Design Space

Activities

• response-surface modeling
• multidimensional assessment
• edge-of-space studies
• scale verification

Deliverables

• Design Space
• PAR
• NOR
• Design Space Verification Report

Phase 5 — PAT Development

Activities

• technology selection
• sensor placement
• sampling strategy
• chemometric model development
• model validation

Deliverables

• PAT Strategy
• PAT URS
• PAT Risk Assessment
• Chemometric Model Report
• PAT Validation/Verification Package

Phase 6 — Control Strategy

Integrate:

CMA Controls + CPP Controls + IPC + PAT + Specifications + Automation + Sampling

Deliverable:

Integrated Pharmaceutical Control Strategy

Phase 7 — Process Qualification

Demonstrate:

• reproducibility;
• robustness;
• equipment capability;
• process capability;
• control-system effectiveness.

Deliverables

• PPQ Protocol
• PPQ Batch Data
• PPQ Report
• Process Validation Summary

Phase 8 — Commercial CPV

Establish:

• data collection;
• SPC;
• capability analysis;
• CQA trending;
• CPP trending;
• PAT monitoring;
• multivariate monitoring;
• annual/product-quality review integration.

Deliverable:

Continued Process Verification Plan and Periodic CPV Report

Phase 9 — Lifecycle Improvement

Use:

CPV → Knowledge → Risk Reassessment → Improvement → Change Control → Regulatory Assessment → Implementation → CPV

This closes the lifecycle loop.

11. Master CQA–CPP–PAT–CPV Matrix

12. Governance Model

A successful QbD program should be cross-functional.

Core Team

• Process Development
• Manufacturing
• QA
• QC
• Analytical Development
• Engineering
• Automation
• Data Science
• Regulatory CMC
• Validation
• MSAT/Technology Transfer
• Supply Chain
• Microbiology, where applicable

Governance Gates

Gate 1: QTPP/CQA approval
Gate 2: Risk assessment approval
Gate 3: DoE strategy approval
Gate 4: Design Space approval
Gate 5: Control Strategy approval
Gate 6: PPQ readiness
Gate 7: CPV readiness
Gate 8: Lifecycle review

13. Minimum Documentation Package

A mature QbD implementation should ultimately contain:

1. QTPP
2. CQA Register
3. CMA Register
4. CPP Register
5. Process Flow Diagram
6. Initial Risk Assessment
7. CQA-CMA-CPP Matrix
8. DoE Protocol
9. DoE Report
10. Statistical Model Report
11. Design Space Report
12. PAR/NOR Justification
13. PAT Strategy
14. PAT URS
15. PAT Risk Assessment
16. Chemometric Model Development Report
17. Chemometric Model Validation Report
18. Control Strategy
19. Process Validation Master Plan/Strategy
20. PPQ Protocol
21. PPQ Report
22. CPV Plan
23. CPV Dashboard
24. CPV Periodic Report
25. Knowledge Management Record
26. Lifecycle Risk Assessment
27. Change Management Strategy
28. Regulatory Impact Assessment

14. Final Integrated Lifecycle Model

The complete framework can be visualized as:

QTPP

↓

CQA Identification

↓

CMA + CPP Identification

↓

QRM / FMEA / Ishikawa

↓

Screening DoE

↓

Optimization DoE

↓

Process Characterization

↓

Design Space

↓

PAT Development

↓

Integrated Control Strategy

↓

Technology Transfer / Scale-Up

↓

Process Qualification

↓

Commercial Manufacturing

↓

CPV + SPC + Multivariate Monitoring

↓

Process Capability Assessment

↓

Knowledge Management

↓

Risk Reassessment

↓

Continuous Improvement / Change Management

↓

ICH Q12 Lifecycle Management

↓

Updated Control Strategy / Design Space / Process Knowledge

↓

Next CPV Cycle

15. Key Regulatory Philosophy

The strongest regulatory position is not simply:

“We tested the process and it passed.”

It is:

“We understand the sources of variability, understand their relationship with CQAs, have demonstrated scientifically justified operating ranges, have implemented an integrated control strategy, continuously monitor process performance, and use lifecycle knowledge to maintain and improve the state of control.”

That philosophy aligns the QbD development approach with FDA’s process-validation lifecycle concept and the ICH Q8/Q9/Q10/Q12 framework. FDA’s Q8/Q9/Q10 Q&A also explicitly describes design-space development, verification, lifecycle management, CPV and continual improvement of the control strategy as connected elements of an enhanced development approach.

Practical Success Criteria

A mature implementation should be able to answer five questions at any time:

1. What can go wrong? → QRM/FMEA
2. Why can it go wrong? → Process understanding/DoE
3. Where can it be controlled? → Design Space/Control Strategy
4. How do we know it remains controlled? → PAT/IPC/CPV/SPC
5. How do we improve it without compromising quality? → Knowledge Management + ICH Q12 Change Management

The ultimate objective is a living, data-driven pharmaceutical process that moves from empirical manufacturing → scientifically understood manufacturing → statistically controlled manufacturing → predictive manufacturing → continuously improving manufacturing.

About the Author

Ramesh Palav is a pharmaceutical manufacturing professional with 21+ years of experience in Oral Solid Dosage manufacturing, production operations, GMP compliance, qualification, validation, QMS and operational excellence. Through Pharma Manufacturing Hub, he shares practical industry knowledge with pharmaceutical professionals, students and manufacturing leaders.