Validate Analytical Method
Formal valid of chromatographic method per ICH Q2(R2) → scope by category → specificity via forced degradation → linearity + range → accuracy + precision → LOD/LOQ + robustness → regulatory compliance.
Use When
- New chromatographic method developed → must validate before routine
- Compendial method verified for suitability in specific lab
- Validated method significant changes → partial|full re-valid
- Prep validation pkg for regulatory submission (NDA, ANDA, MAA, IND)
- Transferring method to new lab|instrument platform
In
Required
- Developed method: Fully optimized + documented (col, mobile phase, gradient, detector, etc.)
- Method category: Assay active, quant impurity, limit impurity, ID test
- Analyte ref stds: Primary refs w/ COA + assigned purity
- Sample matrix: Representative samples + placebo|blank for specificity
Optional
- Regulatory guidance: Beyond ICH Q2 (USP <1225>, FDA, EMA)
- Forced degradation samples: Pre-stressed (acid, base, ox, heat, light) if not prepared
- Validation protocol: Pre-approved w/ acceptance criteria (req in GMP)
- Transfer pkg: If method transfer → originating lab's report
Do
Step 1: Define Scope per ICH Q2(R2)
ID method category + required validation params.
| Parameter |
Cat I: Assay |
Cat II: Impurity (Quant) |
Cat III: Impurity (Limit) |
Cat IV: Identification |
| Specificity |
Yes |
Yes |
Yes |
Yes |
| Linearity |
Yes |
Yes |
No |
No |
| Range |
Yes |
Yes |
No |
No |
| Accuracy |
Yes |
Yes |
No |
No |
| Precision (repeatability) |
Yes |
Yes |
No |
No |
| Precision (intermediate) |
Yes |
Yes |
No |
No |
| LOD |
No |
May be needed |
Yes |
No |
| LOQ |
No |
Yes |
No |
No |
| Robustness |
Yes |
Yes |
Yes |
No |
- Classify method → 4 ICH categories per intended purpose
- Per table, ID required params
- Define acceptance criteria per param before exp work. Typical:
- Linearity: R² >= 0.999 (assay), >= 0.99 (impurity)
- Accuracy: recovery 98.0-102.0% (assay), 80-120% at LOQ
- Repeatability: RSD <= 2.0% (assay), <= 10% at LOQ
- Intermediate: RSD <= 3.0% (assay)
- Draft protocol w/ all params, designs, criteria
- GMP → protocol approval before exp work
Got: Approved protocol w/ method category, required params, designs, pre-defined criteria.
If err: Method category ambiguous (combined assay + impurity) → validate for most stringent. Consult regulatory guidance per submission type.
Step 2: Specificity + Selectivity
- Prep solutions:
- Blank (solvent|diluent)
- Placebo (matrix w/o analyte, e.g., excipients)
- Ref std at working conc
- Spiked placebo (matrix + ref)
- Forced degradation samples (if not avail)
- Forced degradation → potential degradation products:
| Stress Condition |
Typical Treatment |
Target Degradation |
| Acid hydrolysis |
0.1-1 N HCl, 60-80 C, 1-24 h |
5-20% |
| Base hydrolysis |
0.1-1 N NaOH, 60-80 C, 1-24 h |
5-20% |
| Oxidation |
0.3-3% H2O2, RT-60 C, 1-24 h |
5-20% |
| Thermal |
60-80 C, solid state, 1-7 days |
5-20% |
| Photolytic |
ICH Q1B (1.2M lux-hours, 200 Wh/m^2 UV) |
5-20% |
- Inject all + eval:
- No interfering peaks from blank|placebo at analyte RT
- Degradation products resolved from main analyte (Rs >= 1.5)
- Peak purity → DAD spectral purity index|MS
- Mass balance: assay + impurities + degradation = 95-105% initial
- Doc specificity w/ chromatograms from all conditions
Got: Method specific: no interferences, degradation resolved, peak purity confirmed, mass balance 95-105%.
If err: Degradation co-elutes w/ analyte → method not stability-indicating. Return to method dev → improve selectivity (pH, gradient, col chem) before validation.
Step 3: Linearity + Range
- 5+ conc levels spanning intended range:
- Assay: 80-120% of target
- Impurity: LOQ → 120-200% of spec limit
- Dissolution: 10-120% of label claim
- Each conc independent (not serial dilution) → best practice
- Inject each in triplicate (min duplicate)
- Linear regression of response (area|height) vs conc:
- Report slope, intercept, R²
- R² >= 0.999 assay; >= 0.99 impurity
- Eval residual plots:
- Random distribution around zero, no systematic pattern
- Curved → non-linearity → quadratic fit or narrower range
- Calc y-intercept as % of response at 100%:
- <= 2% of 100% response for assay
- Validated range = interval between low + high concs w/ linearity, accuracy, precision
Got: Linear regression R² >= 0.999 (assay)|>= 0.99 (impurity), random residuals, intercept <= 2% target, range defined.
If err: R² below criterion → check prep errors, detector non-linearity (too high conc), analyte instability. Repeat w/ fresh prep. Inherent non-linearity → polynomial calibration|narrow range.
Step 4: Accuracy
- Prep accuracy samples at 3 conc (80%, 100%, 120% target for assay; LOQ, mid, high for impurity)
- Per level → 3 independent replicates (min 9 determinations total)
- Drug substance: compare found vs known (gravimetric)
- Drug product: spiked placebo → add known to placebo + measure recovery
- % recovery per level:
- Recovery (%) = (found / added) x 100
- Acceptance:
| Method Type |
Recovery Range |
RSD at Each Level |
| Assay (drug substance) |
98.0-102.0% |
<= 2.0% |
| Assay (drug product) |
98.0-102.0% |
<= 2.0% |
| Impurity (quantitation) |
80-120% at LOQ, 90-110% at higher levels |
<= 10% at LOQ, <= 5% at higher |
| Cleaning validation |
70-130% (or tighter per company SOP) |
<= 15% |
- Report individual recoveries, mean, RSD per level
Got: Mean recovery within criteria all levels, RSD within limits.
If err: Recovery consistently high|low all levels → systematic err in ref std, sample prep, method (matrix effect → ion suppression in LC-MS). Erratic → investigate prep technique + analyte stability.
Step 5: Precision
3 levels:
- Repeatability (intra-day):
- 1 analyst, 1 instrument, 1 day
- 6 determinations at 100% or 3 levels x 3 replicates (same as accuracy)
- RSD <= 2.0% assay, <= 10% at LOQ for impurity
- Intermediate (inter-day|inter-analyst):
- Repeat repeatability w/ different analyst, day, (if avail) instrument
- Overall RSD combining both
- Overall RSD <= 3.0% assay
- Intermediate >> repeatability → investigate source (analyst, instrument cal, env)
- Reproducibility (method transfer|multi-site):
- Receiving lab → same protocol
- Compare across labs
- F-test (variance) + t-test (mean) or equivalence
| Precision Level |
Design |
Acceptance (Assay) |
Acceptance (Impurity Quant) |
| Repeatability |
n >= 6 at 100%, 1 analyst, 1 day |
RSD <= 2.0% |
RSD <= 10% at LOQ, <= 5% above |
| Intermediate |
2 analysts, 2 days (or 2 instruments) |
RSD <= 3.0% |
RSD <= 15% at LOQ, <= 10% above |
| Reproducibility |
Multi-laboratory |
Per protocol / transfer criteria |
Per protocol / transfer criteria |
Got: Repeatability + intermediate RSDs within criteria. No statistically significant diff between analysts|days|instruments beyond allowed RSD.
If err: Intermediate >> repeatability → ID variable driving variance (analyst prep, ambient temp, instrument cal drift) → control before repeating.
Step 6: LOD, LOQ, Robustness
LOD + LOQ:
- Calc via S/N or std dev approach:
- LOD = 3.3 x (σ / S) → σ = SD of response at low conc, S = slope of cal
- LOQ = 10 x (σ / S)
- Alt: S/N → LOD = S/N >= 3, LOQ = S/N >= 10
- Confirm experimentally → prep solutions at calc'd LOD + LOQ + inject:
- LOD: detectable, not necessarily quantifiable w/ acceptable precision
- LOQ: 6 replicates → RSD <= 10% + accuracy 80-120%
- Report LOD + LOQ w/ method used
Robustness:
- ID critical params (5-7 factors):
- Mobile phase comp (+/- 2% organic)
- MP pH (+/- 0.2 units)
- Col temp (+/- 5 C)
- Flow rate (+/- 10%)
- Detection wavelength (+/- 2 nm)
- Col lot|batch (if avail)
- Vary each within range while holding others (or fractional factorial design for efficiency)
- Eval impact on sys suitability (RT, resolution, tailing, area)
- Params causing sys suitability fail w/in tested range → tightly control + doc as critical
- Summarize robustness in table: each varied param, range tested, impact on responses
Got: LOD + LOQ experimentally confirmed. Robustness done w/ critical params ID'd + control limits.
If err: LOQ precision > 10% RSD → method sensitivity insufficient. Options: increase inj volume, concentrate sample, improve cleanup, more sensitive detector. Param shows not robust (fails SST w/ small variation) → tighten control + flag during transfer.
Check
Traps
- Exp before protocol approval: GMP → data before protocol approval may not be acceptable to regulators. Always approve first.
- Serial dilutions for linearity: Propagates pipetting errs. Each conc independent from common stock for accurate linearity.
- Insufficient forced degradation: < 5% misses important products. > 30% → secondary products complicate. Target 5-20% per condition.
- Confuse repeatability + intermediate: Repeatability = same-day, same-analyst, same-instrument. Intermediate must vary one. Both required Cat I + II.
- Neglect LOQ verify step: Calc LOQ from cal curve insufficient. Must experimentally confirm acceptable precision + accuracy.
- Robustness late: Discovering method not robust after accuracy + precision wastes time + materials. Quick screen early → catch fragile params.
- Incomplete reports: Reviewers expect all raw data, chromatograms (not just tabulated nums), stat analysis, explicit pass/fail per param. Missing → deficiency letters.
→
develop-gc-method — GC method dev preceding validation
develop-hplc-method — HPLC method dev preceding validation
interpret-chromatogram — reading chromatograms during validation
troubleshoot-separation — resolve issues during validation
conduct-gxp-audit — auditing completed validation for GxP
write-standard-operating-procedure — doc validated method as SOP
1---2name: validate-analytical-method-103description: "Validate a chromatographic analytical method per ICH Q2(R2) guidelines: define the validation scope by method category, establish specificity through forced degradation, determine linearity and range, assess accuracy and precision, and establish detection limits and robustness for regulatory submission."4license: MIT5---67# Validate Analytical Method89Formal valid of chromatographic method per ICH Q2(R2) → scope by category → specificity via forced degradation → linearity + range → accuracy + precision → LOD/LOQ + robustness → regulatory compliance.1011## Use When1213- New chromatographic method developed → must validate before routine14- Compendial method verified for suitability in specific lab15- Validated method significant changes → partial|full re-valid16- Prep validation pkg for regulatory submission (NDA, ANDA, MAA, IND)17- Transferring method to new lab|instrument platform1819## In2021### Required2223- **Developed method**: Fully optimized + documented (col, mobile phase, gradient, detector, etc.)24- **Method category**: Assay active, quant impurity, limit impurity, ID test25- **Analyte ref stds**: Primary refs w/ COA + assigned purity26- **Sample matrix**: Representative samples + placebo|blank for specificity2728### Optional2930- **Regulatory guidance**: Beyond ICH Q2 (USP <1225>, FDA, EMA)31- **Forced degradation samples**: Pre-stressed (acid, base, ox, heat, light) if not prepared32- **Validation protocol**: Pre-approved w/ acceptance criteria (req in GMP)33- **Transfer pkg**: If method transfer → originating lab's report3435## Do3637### Step 1: Define Scope per ICH Q2(R2)3839ID method category + required validation params.4041| Parameter | Cat I: Assay | Cat II: Impurity (Quant) | Cat III: Impurity (Limit) | Cat IV: Identification |42|---|---|---|---|---|43| Specificity | Yes | Yes | Yes | Yes |44| Linearity | Yes | Yes | No | No |45| Range | Yes | Yes | No | No |46| Accuracy | Yes | Yes | No | No |47| Precision (repeatability) | Yes | Yes | No | No |48| Precision (intermediate) | Yes | Yes | No | No |49| LOD | No | May be needed | Yes | No |50| LOQ | No | Yes | No | No |51| Robustness | Yes | Yes | Yes | No |52531. Classify method → 4 ICH categories per intended purpose542. Per table, ID required params553. Define acceptance criteria per param before exp work. Typical:56 - Linearity: R² >= 0.999 (assay), >= 0.99 (impurity)57 - Accuracy: recovery 98.0-102.0% (assay), 80-120% at LOQ58 - Repeatability: RSD <= 2.0% (assay), <= 10% at LOQ59 - Intermediate: RSD <= 3.0% (assay)604. Draft protocol w/ all params, designs, criteria615. GMP → protocol approval before exp work6263**Got:** Approved protocol w/ method category, required params, designs, pre-defined criteria.6465**If err:** Method category ambiguous (combined assay + impurity) → validate for most stringent. Consult regulatory guidance per submission type.6667### Step 2: Specificity + Selectivity68691. Prep solutions:70 - Blank (solvent|diluent)71 - Placebo (matrix w/o analyte, e.g., excipients)72 - Ref std at working conc73 - Spiked placebo (matrix + ref)74 - Forced degradation samples (if not avail)752. Forced degradation → potential degradation products:7677| Stress Condition | Typical Treatment | Target Degradation |78|---|---|---|79| Acid hydrolysis | 0.1-1 N HCl, 60-80 C, 1-24 h | 5-20% |80| Base hydrolysis | 0.1-1 N NaOH, 60-80 C, 1-24 h | 5-20% |81| Oxidation | 0.3-3% H2O2, RT-60 C, 1-24 h | 5-20% |82| Thermal | 60-80 C, solid state, 1-7 days | 5-20% |83| Photolytic | ICH Q1B (1.2M lux-hours, 200 Wh/m^2 UV) | 5-20% |84853. Inject all + eval:86 - No interfering peaks from blank|placebo at analyte RT87 - Degradation products resolved from main analyte (Rs >= 1.5)88 - Peak purity → DAD spectral purity index|MS894. Mass balance: assay + impurities + degradation = 95-105% initial905. Doc specificity w/ chromatograms from all conditions9192**Got:** Method specific: no interferences, degradation resolved, peak purity confirmed, mass balance 95-105%.9394**If err:** Degradation co-elutes w/ analyte → method not stability-indicating. Return to method dev → improve selectivity (pH, gradient, col chem) before validation.9596### Step 3: Linearity + Range97981. 5+ conc levels spanning intended range:99 - Assay: 80-120% of target100 - Impurity: LOQ → 120-200% of spec limit101 - Dissolution: 10-120% of label claim1022. Each conc independent (not serial dilution) → best practice1033. Inject each in triplicate (min duplicate)1044. Linear regression of response (area|height) vs conc:105 - Report slope, intercept, R²106 - R² >= 0.999 assay; >= 0.99 impurity1075. Eval residual plots:108 - Random distribution around zero, no systematic pattern109 - Curved → non-linearity → quadratic fit or narrower range1106. Calc y-intercept as % of response at 100%:111 - <= 2% of 100% response for assay1127. Validated range = interval between low + high concs w/ linearity, accuracy, precision113114**Got:** Linear regression R² >= 0.999 (assay)|>= 0.99 (impurity), random residuals, intercept <= 2% target, range defined.115116**If err:** R² below criterion → check prep errors, detector non-linearity (too high conc), analyte instability. Repeat w/ fresh prep. Inherent non-linearity → polynomial calibration|narrow range.117118### Step 4: Accuracy1191201. Prep accuracy samples at 3 conc (80%, 100%, 120% target for assay; LOQ, mid, high for impurity)1212. Per level → 3 independent replicates (min 9 determinations total)1223. Drug substance: compare found vs known (gravimetric)1234. Drug product: spiked placebo → add known to placebo + measure recovery1245. % recovery per level:125 - Recovery (%) = (found / added) x 1001266. Acceptance:127128| Method Type | Recovery Range | RSD at Each Level |129|---|---|---|130| Assay (drug substance) | 98.0-102.0% | <= 2.0% |131| Assay (drug product) | 98.0-102.0% | <= 2.0% |132| Impurity (quantitation) | 80-120% at LOQ, 90-110% at higher levels | <= 10% at LOQ, <= 5% at higher |133| Cleaning validation | 70-130% (or tighter per company SOP) | <= 15% |1341357. Report individual recoveries, mean, RSD per level136137**Got:** Mean recovery within criteria all levels, RSD within limits.138139**If err:** Recovery consistently high|low all levels → systematic err in ref std, sample prep, method (matrix effect → ion suppression in LC-MS). Erratic → investigate prep technique + analyte stability.140141### Step 5: Precision1421433 levels:1441451. **Repeatability (intra-day)**:146 - 1 analyst, 1 instrument, 1 day147 - 6 determinations at 100% or 3 levels x 3 replicates (same as accuracy)148 - RSD <= 2.0% assay, <= 10% at LOQ for impurity1492. **Intermediate (inter-day|inter-analyst)**:150 - Repeat repeatability w/ different analyst, day, (if avail) instrument151 - Overall RSD combining both152 - Overall RSD <= 3.0% assay153 - Intermediate >> repeatability → investigate source (analyst, instrument cal, env)1543. **Reproducibility** (method transfer|multi-site):155 - Receiving lab → same protocol156 - Compare across labs157 - F-test (variance) + t-test (mean) or equivalence158159| Precision Level | Design | Acceptance (Assay) | Acceptance (Impurity Quant) |160|---|---|---|---|161| Repeatability | n >= 6 at 100%, 1 analyst, 1 day | RSD <= 2.0% | RSD <= 10% at LOQ, <= 5% above |162| Intermediate | 2 analysts, 2 days (or 2 instruments) | RSD <= 3.0% | RSD <= 15% at LOQ, <= 10% above |163| Reproducibility | Multi-laboratory | Per protocol / transfer criteria | Per protocol / transfer criteria |164165**Got:** Repeatability + intermediate RSDs within criteria. No statistically significant diff between analysts|days|instruments beyond allowed RSD.166167**If err:** Intermediate >> repeatability → ID variable driving variance (analyst prep, ambient temp, instrument cal drift) → control before repeating.168169### Step 6: LOD, LOQ, Robustness170171**LOD + LOQ**:1721731. Calc via S/N or std dev approach:174 - LOD = 3.3 x (σ / S) → σ = SD of response at low conc, S = slope of cal175 - LOQ = 10 x (σ / S)176 - Alt: S/N → LOD = S/N >= 3, LOQ = S/N >= 101772. Confirm experimentally → prep solutions at calc'd LOD + LOQ + inject:178 - LOD: detectable, not necessarily quantifiable w/ acceptable precision179 - LOQ: 6 replicates → RSD <= 10% + accuracy 80-120%1803. Report LOD + LOQ w/ method used181182**Robustness**:1831844. ID critical params (5-7 factors):185 - Mobile phase comp (+/- 2% organic)186 - MP pH (+/- 0.2 units)187 - Col temp (+/- 5 C)188 - Flow rate (+/- 10%)189 - Detection wavelength (+/- 2 nm)190 - Col lot|batch (if avail)1915. Vary each within range while holding others (or fractional factorial design for efficiency)1926. Eval impact on sys suitability (RT, resolution, tailing, area)1937. Params causing sys suitability fail w/in tested range → tightly control + doc as critical1948. Summarize robustness in table: each varied param, range tested, impact on responses195196**Got:** LOD + LOQ experimentally confirmed. Robustness done w/ critical params ID'd + control limits.197198**If err:** LOQ precision > 10% RSD → method sensitivity insufficient. Options: increase inj volume, concentrate sample, improve cleanup, more sensitive detector. Param shows not robust (fails SST w/ small variation) → tighten control + flag during transfer.199200## Check201202- [ ] Method category ID'd + all params per ICH Q2(R2)203- [ ] Validation protocol w/ pre-defined criteria204- [ ] Specificity demonstrated: no interferences, degradation resolved, peak purity205- [ ] Mass balance 95-105% for forced degradation206- [ ] Linearity R² >= 0.999 (assay)|>= 0.99 (impurity), residuals random207- [ ] Accuracy at 3 levels, recovery within criteria208- [ ] Repeatability RSD within limits (<= 2.0% assay)209- [ ] Intermediate RSD within limits (<= 3.0% assay)210- [ ] LOD + LOQ experimentally confirmed (LOQ precision <= 10% RSD)211- [ ] Robustness w/ critical params ID'd212- [ ] All raw data, calc, chromatograms in validation report213214## Traps215216- **Exp before protocol approval**: GMP → data before protocol approval may not be acceptable to regulators. Always approve first.217- **Serial dilutions for linearity**: Propagates pipetting errs. Each conc independent from common stock for accurate linearity.218- **Insufficient forced degradation**: < 5% misses important products. > 30% → secondary products complicate. Target 5-20% per condition.219- **Confuse repeatability + intermediate**: Repeatability = same-day, same-analyst, same-instrument. Intermediate must vary one. Both required Cat I + II.220- **Neglect LOQ verify step**: Calc LOQ from cal curve insufficient. Must experimentally confirm acceptable precision + accuracy.221- **Robustness late**: Discovering method not robust after accuracy + precision wastes time + materials. Quick screen early → catch fragile params.222- **Incomplete reports**: Reviewers expect all raw data, chromatograms (not just tabulated nums), stat analysis, explicit pass/fail per param. Missing → deficiency letters.223224## →225226- `develop-gc-method` — GC method dev preceding validation227- `develop-hplc-method` — HPLC method dev preceding validation228- `interpret-chromatogram` — reading chromatograms during validation229- `troubleshoot-separation` — resolve issues during validation230- `conduct-gxp-audit` — auditing completed validation for GxP231- `write-standard-operating-procedure` — doc validated method as SOP